This page contains one-sentence summaries of ICML 2024 accepted papers, generated by the compressor, my personal LLM-based project.
https://openreview.net/forum?id=zxxSJAVQPc
Compressor summary: GraphsGPT uses a Graph2Seq encoder to transform Non-Euclidean graphs into learnable Graph Words, and a GraphGPT decoder to reconstruct the original graph, achieving state-of-the-art results in graph representation learning and generation.
https://openreview.net/forum?id=zwUEk9WpsR
Compressor summary: The text discusses how server-assisted federated learning (SA-FL) can address incomplete client participation and provide theoretical and practical benefits over conventional FL.
https://openreview.net/forum?id=ztn8FCR1td
Compressor summary: The paper presents a new architecture (Mamba-2) that combines state-space models and attention, improving the speed and performance of language modeling.
https://openreview.net/forum?id=zrQIc9mQQN
Compressor summary: The paper proposes a new method to improve graph neural networks for node classification by modeling the joint distribution of nodes and their clusters, enhancing accuracy and robustness against adversarial attacks.
https://openreview.net/forum?id=znz261CQK7
Compressor summary: Key points: - The paper studies random and SGD-optimized neural networks with zero training error - Overparameterization in terms of width helps generalization due to SGD bias - Overparameterization in terms of depth harms generalization but random and SGD-optimized networks behave similarly, suggesting an architectural bias Summary: The paper investigates how optimization and architecture affect generalization of neural networks with zero training error, finding that overparameterization in width is good for generalization because of SGD bias, while overparameterization in depth is bad for generalization but random and SGD-optimized networks are similar, indicating an architectural bias.
https://openreview.net/forum?id=znKAWRZSF9
Compressor summary: S3GCL is a new graph representation learning method that addresses challenges of homophily assumptions and scalability by using cosine-parameterized Chebyshev polynomials as filters and an MLP encoder with positive pairs for context awareness.
https://openreview.net/forum?id=zkjGpZrIX3
Compressor summary: The text introduces Trustworthy Actionable Perturbations (TAP), a framework to modify inputs that change true class probabilities instead of fooling classifiers, with verification, cost, reward, and goal definitions for real-world applications.
https://openreview.net/forum?id=zkcya47Sq5
Compressor summary: To compare two binary classifiers, spend the budget on collecting a single noisy label for more samples rather than aggregating multiple labels via majority vote, as this increases accuracy and provides better sample size bounds.
https://openreview.net/forum?id=zji9DLksTz
Compressor summary: FRB is a method to improve binary image super-resolution by using second-order residual binarization and distillation-guided binarization training, achieving better results than existing methods.
https://openreview.net/forum?id=zj7YuTE4t8
Compressor summary: The paper proposes a method where multiple language models debate and improve each other's responses, enhancing reasoning and validity of generated content.
https://openreview.net/forum?id=zgiT3uxvCF
Compressor summary: The paper proposes ccGPFA, a conditionally-conjugate method for modeling neural activity from spike count data that enables tractable inference using variational EM and sparse Gaussian Processes.
https://openreview.net/forum?id=zfmwAaB9Nw
Compressor summary: The paper proposes a new technique to improve privacy analysis and algorithms for combinatorial optimization problems and heavy hitter identification in data streams.
https://openreview.net/forum?id=zdNTiTs5gU
Compressor summary: Deep heteroscedastic regression improves covariance estimation using gradient and curvature information, and introduces a metric to evaluate accuracy without supervision.
https://openreview.net/forum?id=zcIV8OQFVF
Compressor summary: The paper investigates reward hacking in Reinforcement Learning from Human Feedback using Long Short-Term Memories (LLMs) and proposes an improved evaluation protocol and reward model to address this issue.
https://openreview.net/forum?id=zc3bAEI5lp
Compressor summary: The paper proposes a novel method using sum-product networks to create a single model that performs both privacy-preserving classification and data generation, outperforming existing approaches in stability and utility.
https://openreview.net/forum?id=zatLnLvbs8
Compressor summary: ConPaS is a new machine learning framework that predicts and fixes integer variable assignments to solve MILPs faster and better than existing methods.
https://openreview.net/forum?id=zajsXCxMgW
Compressor summary: The paper introduces a new method for distributional reinforcement learning that separates transition structure and reward, using a distributional successor measure that describes the consequences of behavior, which can be learned from data and enables zero-shot risk-sensitive policy evaluation.
https://openreview.net/forum?id=zWIS8I9G9B
Compressor summary: The paper studies online LQR control with heterogeneous cost curvatures and provides a novel analysis using Newton decrement to improve adaptivity and performance.
https://openreview.net/forum?id=zS8zUuAU8T
Compressor summary: The paper proposes a new framework that uses distillation and target-relevant information to improve domain adaptive object detection, addressing source bias and achieving better classification and localization in both domains.
https://openreview.net/forum?id=zRrzSLwNHQ
Compressor summary: The paper proposes a new method for graph neural networks that counts all structures in a pattern basis, improving their expressive power and counting abilities without increasing complexity.
https://openreview.net/forum?id=zMwFvxr6CV
Compressor summary: The paper proposes an agent that generates instructions to improve the zero-shot reasoning abilities of large language models on various tasks and datasets, achieving state-of-the-art performance.
https://openreview.net/forum?id=zMue490KMr
Compressor summary: Grokking is a widespread phenomenon in deep neural networks where generalization occurs long after zero training error, and it leads to delayed robustness against adversarial examples due to the linearization of the network mapping.
https://openreview.net/forum?id=zMsMQJraEj
Compressor summary: The paper studies how two learning agents interact over time, shows that existing benchmarks don't capture their dynamics well, and proposes new algorithms and environments to improve their performance.
https://openreview.net/forum?id=zMGUDsPopK
Compressor summary: The paper proposes a new method called Distance Aware Bottleneck that enhances deep neural networks with uncertainty estimation and out-of-distribution detection by learning a codebook of input representations.
https://openreview.net/forum?id=zL9q2JD1dC
Compressor summary: GALA3D is a framework that uses language models to generate realistic 3D scenes with controllable editing and consistent properties.
https://openreview.net/forum?id=zII3Olw7cr
Compressor summary: The paper presents TULIP, a method to improve single-pass uncertainty estimation in neural networks by preserving feature distances at intermediate layers.
https://openreview.net/forum?id=zFHaB7KESM
Compressor summary: The paper proposes a method to learn general nonlinear representations from multiple data sources with different distributions and dependencies, providing statistical guarantees and requiring fewer samples than existing methods.
https://openreview.net/forum?id=zEqeNEuiJr
Compressor summary: The paper proposes signSGD-FD, a distributed learning optimizer that uses federated defense to handle both honest and adversarial workers, ensuring fast convergence even under attacks.
https://openreview.net/forum?id=zDCwJQY3eI
Compressor summary: The paper proposes a practical implementation of the polar factorization theorem for vector fields, using neural networks and optimal transport to parameterize the components, and explores its applications in machine learning.
https://openreview.net/forum?id=zCmMkWK4Ly
Compressor summary: The paper proposes ICES, a method to encourage multi-agent exploration in sparse reward environments by assessing each agent's contribution and using global transition information during training.
https://openreview.net/forum?id=zB6VQzDmK8
Compressor summary: The paper proposes a method to obtain realistic decisions from ReLU networks by modeling the data manifold as an optimization constraint and using adaptive sampling to solve the problem efficiently.
https://openreview.net/forum?id=z8sYc334fU
Compressor summary: The study examines the behavior, representativeness, and point-wise information content of dataset distillation, finding that it retains high task performance by compressing early training dynamics and contains meaningful semantic information.
https://openreview.net/forum?id=z7zHsNFXHc
Compressor summary: ResNet can efficiently approximate functions such as polynomials and smooth ones with fewer tunable weights than ReLU networks, achieving optimal approximation rates.
https://openreview.net/forum?id=z5Ux2u6t7U
Compressor summary: DITTO is a framework to control text-to-music diffusion models without fine-tuning by optimizing initial noise latents and achieving high quality, flexible, and efficient music generation.
https://openreview.net/forum?id=z3PUNzdmGs
Compressor summary: The paper presents a new algorithm for maintaining a randomized mapping of a weighted graph to an $\ell_p$ space while preserving the distance between vertices, with low expected distortion and fast update time, even in dynamic settings where edge weights change over time.
https://openreview.net/forum?id=z373OXJXWU
Compressor summary: This paper analyzes the convergence properties of doubly stochastic gradient descent (doubly SGD), a popular optimization strategy for problems with intractable expectations, and shows that random reshuffling can improve its performance.
https://openreview.net/forum?id=yzNEkTmcoF
Compressor summary: The paper proposes and tests a new estimator called 'triple changes' that allows for a more nuanced assessment of causal effects by incorporating information from potential outcomes, compared to existing methods like difference-in-differences or triple differences.
https://openreview.net/forum?id=yyYMAprcAR
Compressor summary: Tied input-output embeddings in language models relate to Harris' distributional hypothesis and contextual similarity of words.
https://openreview.net/forum?id=ytz2naZoDB
Compressor summary: Key points: - The paper proposes a framework to train SDEs with policy gradients for generating samples with high rewards - The framework constrains the SDE to be consistent with its associated perturbation process, which covers the entire space and is easy to sample - The method is applied to structure-based drug design and improves the binding affinity of ligand molecules Summary: The paper presents a policy gradient framework for training SDEs that are compatible with their perturbation processes, enabling effective and efficient generation of high-reward samples. The method is used for structure-based drug design and achieves the best score on a dataset.
https://openreview.net/forum?id=yrFUJzcTsk
Compressor summary: HesScale is a fast and stable approximation of Hessian diagonals that outperforms existing methods in small networks for second-order optimization and reinforcement learning problems.
https://openreview.net/forum?id=yoqdlynCRs
Compressor summary: This paper studies how to adjust the size of experts in Mixture of Experts (MoE) models for efficient language processing, introducing a new hyperparameter called granularity and showing how it affects scaling laws.
https://openreview.net/forum?id=yoTCwNqQS6
Compressor summary: The paper proposes a new method called Label-Encoding Risk Minimization (LERM) that uses unlabeled samples to improve the learning of labeled samples and balance prediction discriminability and diversity in scenarios with insufficient labels.
https://openreview.net/forum?id=yo9Jyt3XCY
Compressor summary: The paper examines how social media influencers boost the visibility and citations of AI and ML papers, and suggests they should promote diversity in their content.
https://openreview.net/forum?id=ymgcTqrZLT
Compressor summary: The paper proposes a new method to find an average distribution from multiple probability measures using neural networks and optimal transport, with theoretical guarantees and applications to image data.
https://openreview.net/forum?id=ykgZk6vFrh
Compressor summary: This paper studies a repeated game where a principal tries to learn how to incentivize an agent to act in the principal's best interest in various settings, motivated by applications like healthcare and ecology.
https://openreview.net/forum?id=ykZYLBcA9g
Compressor summary: The paper proposes a new consistent method for complementary-label learning that does not rely on the uniform distribution assumption or ordinary-label training sets, and shows its effectiveness in experiments.
https://openreview.net/forum?id=ykRY34kL3j
Compressor summary: The paper proposes a new method called BeCa, which combines generative and contrastive methods for self-supervised full-face gaze estimation and improves performance over existing approaches on various datasets.
https://openreview.net/forum?id=yj8h567Ia7
Compressor summary: The paper proposes a method to speed up pragmatic program synthesis by using a global ranking derived from partial rankings of programs generated by an exact RSA synthesizer.
https://openreview.net/forum?id=yhpDKSw7yA
Compressor summary: This paper proposes a framework for policy optimization with random preference flips, introduces a novel loss function to debias the effect of noise, and proves its sub-optimality bound in theory.
https://openreview.net/forum?id=yh6Y7ppf46
Compressor summary: The paper proposes a novel method to combine machine learning and legacy numerical codes without modification for faster scientific computing.
https://openreview.net/forum?id=ycXo4tQIpN
Compressor summary: The paper proposes a method to learn shadow variables from observational data for estimating treatment effects in the presence of collider bias, using hypothesis testing and a novel treatment effect estimator.
https://openreview.net/forum?id=ycLHJuLYuD
Compressor summary: SAFECLIP is a defense method for pre-training CLIP models against targeted data poisoning and backdoor attacks by using unimodal contrastive learning and dividing data into safe and risky sets.
https://openreview.net/forum?id=yb5xV8LFDq
Compressor summary: The paper proposes a new method for selecting small coresets that maintain model performance while minimizing the coreset size in deep learning algorithms, and provides theoretical and empirical results to support its effectiveness.
https://openreview.net/forum?id=yY6N89IlHa
Compressor summary: The paper introduces CLIF, a new neuron model for spiking neural networks that improves accuracy by facilitating the backpropagation of temporal gradients, and shows its superior performance over other models and conventional ANNs on various datasets.
https://openreview.net/forum?id=yXlQL9goY8
Compressor summary: The paper develops new information-theoretic methods to analyze and compare different types of contrastive learning algorithms, including pointwise, pairwise, triplet, quadruplet, and higher-order scenarios.
https://openreview.net/forum?id=yUxdk32TU6
Compressor summary: The paper presents a novel framework for generating diverse and controllable adversarial attacks on large language models using an efficient text generation algorithm.
https://openreview.net/forum?id=yUPBkPKzHw
Compressor summary: ELBERT is a long-term fairness concept for machine learning models in sequential decision-making that reduces biases and maintains high utility.
https://openreview.net/forum?id=yTz0u4B8ug
Compressor summary: Memoria is a novel artificial neural network memory system inspired by human memory that outperforms conventional techniques in various tasks.
https://openreview.net/forum?id=yTXv8KDD1P
Compressor summary: The paper proposes tight non-vacuous bounds on the greedy heuristic InfoMax for binary test prediction, assuming that the conditional probability of a test being 'true' is within $\delta$ units of one-half, and analyzes two scenarios with modest values of $\delta$.
https://openreview.net/forum?id=yShA4VPYZB
Compressor summary: This paper explores Euclidean symmetries in cooperative multi-agent reinforcement learning, designing neural networks with symmetric constraints that improve performance and generalization in various applications.
https://openreview.net/forum?id=yQfA0etfB7
Compressor summary: The text studies algorithms for approximating subspaces of points in the $\ell_p$ norm, giving fast and strong coreset constructions that can be applied to various geometric problems.
https://openreview.net/forum?id=yPDTXQwUPy
Compressor summary: The paper introduces *ETHER*, a new method to fine-tune foundation models with fewer parameters, less performance deterioration, and hyperparameter robustness.
https://openreview.net/forum?id=yOe5lqDPvM
Compressor summary: RODEO is a data-centric approach that uses text-to-image models to generate diverse and near-distribution outliers for robust outlier detection in adversarial settings.
https://openreview.net/forum?id=yL6hljtjW4
Compressor summary: The paper proposes a method called STATA that improves the efficiency of training Spiking Transformers by sparsifying tokens, identifying important ones across timesteps, and aligning attention maps.
https://openreview.net/forum?id=yHs3jIPgaF
Compressor summary: The paper proposes a framework to study social prediction without assuming convexity, known mapping, or first-order information, by reparameterizing the objective function and optimizing it iteratively.
https://openreview.net/forum?id=yHRxnhKyEJ
Compressor summary: This paper investigates the effect of local steps in Federated Learning on feature learning and generalization, and shows that they help improve performance and reduce communication costs.
https://openreview.net/forum?id=yFUdZfbEme
Compressor summary: The Imprecise Domain Generalisation framework helps machine learners deal with uncertainty in out-of-distribution generalisation by allowing them to optimise against a range of strategies during training and enabling operators to specify their preferences at deployment time.
https://openreview.net/forum?id=yDXnXJE1RK
Compressor summary: The paper proposes VP G-CNNs, a novel approach to capture varying levels of partial equivariance for different data instances using variational inference and redesigned distributions.
https://openreview.net/forum?id=y8YovS0lOg
Compressor summary: The paper investigates how different optimization algorithms like RMSProp and Adam regularize solutions during training by analyzing their corresponding ordinary differential equations and their dependence on hyperparameters.
https://openreview.net/forum?id=y8NevOhrnW
Compressor summary: The paper investigates neural collapse in multi-label classification and shows how it affects feature representation and optimizer behavior, leading to improved performance.
https://openreview.net/forum?id=y6y2HauOpR
Compressor summary: The paper studies robust Markov games with rectangular uncertainty, shows a connection between robust Nash equilibrium and regularized Markov games, provides a planning algorithm and provable guarantees, and identifies a special class of games that can be solved efficiently for reward-uncertain two-player zero-sum cases.
https://openreview.net/forum?id=y5L8W0KRUX
Compressor summary: Evolutionary Ranking (EvoRank) is a new training objective for AI-based protein engineering frameworks that uses evolutionary information from multiple sequence alignments to improve mutation effect predictions and learn diverse protein representations.
https://openreview.net/forum?id=xzX7kf486K
Compressor summary: Neural Diffusion Models generalize conventional diffusion models by enabling non-linear transformations of data, improving generative tasks performance and sample quality.
https://openreview.net/forum?id=xye7iNsgXn
Compressor summary: Key points: - Large-scale recommendation systems need to handle high cardinality, heterogeneous features and billions of user actions - DLRMs fail to scale with compute, while Transformers succeed in language and vision domains - HSTU is a new architecture for streaming recommendation data that outperforms baselines and is faster than FlashAttention2-based Transformers - Generative Recommenders with HSTU improve metrics in online A/B tests and reduce carbon footprint Summary: The authors propose HSTU, a new Transformer-based architecture for recommendation systems that handles high features, scales well with compute, and improves both performance and sustainability.
https://openreview.net/forum?id=xxL7CEWuxz
Compressor summary: The paper explores how intrinsic dimension can be used as a metric to measure and improve the prunability of vision-language models by modality, finding that visual representations are more crucial while language representations are more robust.
https://openreview.net/forum?id=xwxUbBHC1q
Compressor summary: The paper analyzes the convergence properties and theoretical guarantees of DeepWalk algorithm on graphs generated by Stochastic Block Model, a simple model to study algorithm behavior on large graphs.
https://openreview.net/forum?id=xwOENWCo46
Compressor summary: The paper proposes a method to learn diverse graph representations by approximating graph entropy using orthonormal neural networks, which improve performance in unsupervised and semi-supervised learning tasks.
https://openreview.net/forum?id=xuX2rDSSco
Compressor summary: GEAM is a molecular generative framework for drug discovery that considers target chemical properties and updates its fragment vocabulary dynamically during the generation process.
https://openreview.net/forum?id=xtwCf7iAs2
Compressor summary: The paper introduces CMANPs, a variant of Neural Processes that uses constant memory and attention blocks to model predictive uncertainty efficiently.
https://openreview.net/forum?id=xtKWwB6lzT
Compressor summary: The paper critiques offline reinforcement learning in dynamic treatment regimes, citing concerns about evaluation metrics, baselines, and RL formulations; it also presents a case study showing RL performance variations and the possibility of random baselines outperforming RL algorithms.
https://openreview.net/forum?id=xqqccG7gf1
Compressor summary: The text describes a privacy vulnerability of diffusion models in image synthesis and proposes an improved membership inference attack that uses quantile regression and bootstrapping to detect if a given example belongs to the training data or not with higher accuracy and lower computational cost than previous methods.
https://openreview.net/forum?id=xpSlt67vxQ
Compressor summary: R-Bench is a benchmark to evaluate vision relationship hallucinations in large language models by testing their understanding of visual relationships and content.
https://openreview.net/forum?id=xnQ1qoly7Q
Compressor summary: The paper introduces RoboCodeX, a framework for generating detailed robotic actions from high-level human instructions using tree-structured multimodal code generation and pre-training with a specialized dataset.
https://openreview.net/forum?id=xm2lU7tteQ
Compressor summary: The paper explores how Transformer models with a nonlinear representation layer can learn effectively in context, and proves that their loss landscape is benign and stable under certain conditions.
https://openreview.net/forum?id=xlr6AUDuJz
Compressor summary: The authors introduce the WMDP benchmark to measure hazardous knowledge in LLMs and propose RMU, an unlearning method that reduces malicious use while preserving general capabilities.
https://openreview.net/forum?id=xlWcdtCyOC
Compressor summary: This paper presents InstructSpeech, a multi-task language model that can edit speech based on natural language instructions, achieving state-of-the-art results in eleven tasks.
https://openreview.net/forum?id=xl82CcbYaT
Compressor summary: The paper shows that simple linear models and their variants perform well in time series forecasting and can be interpreted as unconstrained linear regression over expanded features, leading to better forecasts.
https://openreview.net/forum?id=xl2yU3dsHK
Compressor summary: The paper presents new differentially private algorithms for online convex optimization that work well with non-smooth and high-dimensional loss functions, using sampling from log-concave densities and rejection sampling.
https://openreview.net/forum?id=xgoilgLPGD
Compressor summary: The paper introduces Langevin Actor-Critic (LAC), a method that combines sampling-based Langevin policy with optimization-based actor-critic to enable safe reinforcement learning in complex tasks.
https://openreview.net/forum?id=xcyKKACmSd
Compressor summary: S3O is a novel method that learns parametric models for shape and motion from monocular videos without requiring additional annotations or extensive computational resources, improving 3D reconstruction accuracy and robustness.
https://openreview.net/forum?id=xcDRx8vzCa
Compressor summary: CHAI reduces memory and compute requirements for large language models by clustering correlated attention heads at runtime.
https://openreview.net/forum?id=xbQqhojHTg
Compressor summary: PULCPBF is a two-stage method for PU learning that uses weak classifiers as a probability boundary fence and self-training to improve performance.
https://openreview.net/forum?id=xaSpuvNYwS
Compressor summary: Key points: - Diffusion models can improve adversarial robustness but have limitations - Robust Diffusion Classifier (RDC) is a generative classifier based on a pre-trained diffusion model - RDC uses multi-head diffusion and efficient sampling strategies to reduce computational cost - RDC outperforms adversarial training models against various adaptive attacks on CIFAR-10 Summary: The paper proposes Robust Diffusion Classifier, a generative classifier that leverages a pre-trained diffusion model to achieve better adversarial robustness than existing methods.
https://openreview.net/forum?id=xZO7SmM12y
Compressor summary: The paper proposes EOE, a method that uses large language models to generate outlier class labels and a score function based on potential outlier penalty for zero-shot OOD detection in open-world scenarios.
https://openreview.net/forum?id=xWI0MKwJSS
Compressor summary: Shuffling and Poisson subsampling are two different batch sampling methods for private machine learning that have significantly different privacy guarantees, and practitioners should be careful when reporting the privacy analysis of private SGD.
https://openreview.net/forum?id=xW79geE0RA
Compressor summary: DLPA is a new RL algorithm for PAMDPs that learns a dynamics model, plans with a modified control method, and outperforms existing methods in sample efficiency and performance.
https://openreview.net/forum?id=xVXnXk9I3I
Compressor summary: The paper proposes a new method for aligning text-to-image diffusion models with user preferences by using dense rewards and temporal discounting, which improves the efficiency and effectiveness of preference alignment.
https://openreview.net/forum?id=xTYIAD2NND
Compressor summary: The paper proposes a formal framework to analyze generalization in dynamical systems reconstruction from time series data, showing that black-box deep learning techniques often fail to achieve it and suggesting ways to improve it.
https://openreview.net/forum?id=xSkIxKdO08
Compressor summary: Key points: - Optimization layers in deep neural networks improve structured learning but lack interpretability - Counterfactual explanations can make them more transparent by providing alternative outcomes - Variational autoencodters enable counterfactuals in latent space with plausibility - CF-OPT is a first-order optimization algorithm for finding counterfactual explanations in structured learning Summary: The paper proposes a novel algorithm (CF-OPT) that uses variational autoencoders to find interpretable counterfactual explanations for optimization layers in deep neural networks.
https://openreview.net/forum?id=xSizvCoI79
Compressor summary: The paper proposes Subgraph-To-Node translation, a method for learning subgraph representations that reduces memory and computational costs and performs better than existing graph neural networks.
https://openreview.net/forum?id=xS2YKQlBIZ
Compressor summary: PRGD is a novel algorithm that maximizes the margin at an exponential rate for linearly separable data, while existing algorithms like GD and NGD fail under certain conditions and only achieve a polynomial rate.
https://openreview.net/forum?id=xQiYCmDrjp
Compressor summary: The paper proposes a novel temporal distance for stochastic settings that satisfies the triangle inequality, enabling better planning and control in reinforcement learning.
https://openreview.net/forum?id=xPypr0kufs
Compressor summary: The paper proposes using Fusion Frames to quantize Transformers to two bits, achieving efficient and accurate models with known recovery guarantees and denoising filters.
https://openreview.net/forum?id=xPmSNLle1w
Compressor summary: The paper proposes an alternative pruning test method for $\ell_0$-regularized problems that improves the solving time of Branch-and-Bound algorithms in machine-learning applications.
https://openreview.net/forum?id=xMJT4XW468
Compressor summary: The paper develops a theory of network kernels in deep neural networks that explains how their Bayesian prior adapts to data and features using large deviation and field-theoretic approaches.
https://openreview.net/forum?id=xLikRS9OhW
Compressor summary: The paper investigates the reasoning capabilities of Sparse Transformer and Linear Transformer, finding that they are expressive enough for Dynamic Programming problems but require more model size than standard Transformer, and identifies a class of problems where they are more efficient.
https://openreview.net/forum?id=xJUhgvM2u8
Compressor summary: Graph neural stochastic diffusion (GNSD) is a new method for estimating uncertainty in graph neural network predictions by connecting them to stochastic partial differential equations with a $Q$-Wiener process and two networks to ensure accurate prediction and uncertainty propagation.
https://openreview.net/forum?id=xJMZbdiQnf
Compressor summary: The proposed method, LESR, uses a large language model to generate task-related state representation codes for reinforcement learning, improving sample efficiency and performance in Mujoco and Gym-Robotics tasks.
https://openreview.net/forum?id=xIRKB5nRJl
Compressor summary: The authors propose a framework to train robots using multimodal prompts, combining vision and language signals, and achieve state-of-the-art performance on robot manipulation tasks.
https://openreview.net/forum?id=xGlVkBSDdt
Compressor summary: The authors propose a new method to learn intensity functions of counting processes using neural controlled differential equations and signature-based estimation, with theoretical guarantees and applications in various domains.
https://openreview.net/forum?id=xFk0w9zoV3
Compressor summary: EE-LLM is a framework that trains and uses large language models with early exiting to speed up inference while maintaining quality.
https://openreview.net/forum?id=xFDJBzPhci
Compressor summary: The paper proposes a new objective function to improve out-of-distribution generalization and detection in vision-language pre-trained models during fine-tuning, by minimizing the gradient magnitude of energy scores on training data.
https://openreview.net/forum?id=xFCA2yWVs4
Compressor summary: The paper introduces a method to train Markov chains with neural networks for efficient and well-mixed sampling, using involutive Metropolis-Hastings kernels from reversible neural networks for detailed balance.
https://openreview.net/forum?id=xF656w37Mj
Compressor summary: The paper explores using uncertainty quantification (UQ) predictions in online algorithms for ski rental and online search problems, and proposes a new online learning framework to leverage UQ effectively.
https://openreview.net/forum?id=xEB2oF3vvb
Compressor summary: The paper advocates for application-driven research in machine learning as a way to create impactful solutions and foster innovation, while suggesting reforms in the current reviewing, hiring, and teaching practices.
https://openreview.net/forum?id=xC7SYAZygF
Compressor summary: SBMI is a method that uses neural networks and simulations to systematically select model components in composite scientific models, enabling data-driven discovery and uncertainty-informed decision making.
https://openreview.net/forum?id=xB6YJZOKyT
Compressor summary: The paper proposes RVI-SAC, an off-policy DRL method using average reward criterion, which performs well on Mujoco locomotion tasks.
https://openreview.net/forum?id=x2zxPwCkAZ
Compressor summary: FedBAT is a novel framework for federated learning that learns binary model updates during local training, reducing approximation errors and improving accuracy.
https://openreview.net/forum?id=x1G7ieRgRd
Compressor summary: The paper proposes a novel privacy accounting method and sparsification scheme for mean estimation in central differential privacy that improves MSEs, communication efficiency, and compatibility with streaming differential privacy and DP-FTRL optimizers.
https://openreview.net/forum?id=x0yIaw2fgk
Compressor summary: The paper investigates how observation and reward modeling tasks in world models affect sample-efficient MBRL and proposes HarmonyDream, a method that balances these two tasks to improve performance.
https://openreview.net/forum?id=x0vLj1S6Wg
Compressor summary: The paper introduces new partial monitoring strategies using randomized confidence bounds and extends regret guarantees to stochastic settings with side information.
https://openreview.net/forum?id=wwItuHdus6
Compressor summary: The text proposes a deep learning framework that simplifies solving various optimal transport problems without needing optimal couplings or simulating dynamics.
https://openreview.net/forum?id=wuQ2DRPAuy
Compressor summary: Robust-HDP improves federated learning utility and convergence speed by efficiently estimating and reducing differential privacy noise across heterogeneous clients.
https://openreview.net/forum?id=wrTzLoqbCg
Compressor summary: TimeSiam is a self-supervised pre-training framework for time series that leverages Siamese networks to capture temporal correlations and outperforms other methods in forecasting and classification tasks.
https://openreview.net/forum?id=wnkC5T11Z9
Compressor summary: The paper explores how attention mechanisms in transformer models can offer insights on their behavior and compares them with post-hoc explanations.
https://openreview.net/forum?id=wmljUnbjy6
Compressor summary: The paper introduces simplicial scattering networks (SSNs), a parameter-free model that extracts task-agnostic features from simplicial complex data without labels using random walk matrices, improving robustness and performance in various tasks.
https://openreview.net/forum?id=wleAlsklEh
Compressor summary: Matrix-SSL is a novel method that leverages matrix information theory to improve non-contrastive learning methods by aligning covariance matrices and achieving better results than previous state-of-the-art methods on image and language tasks.
https://openreview.net/forum?id=wlOaG9g0uq
Compressor summary: This paper proposes three approaches to understand and manipulate emotions in generative AI models using psychological theories, and shows that they can comprehend emotional stimuli like the human brain's dopamine mechanism.
https://openreview.net/forum?id=wlBtHP8KqS
Compressor summary: The paper proposes a user-level differential privacy method for sparse linear regression that eliminates the dimension dependency and performs better than item-level methods with the same number of samples.
https://openreview.net/forum?id=wkCUmO7oi2
Compressor summary: JoCo is a novel framework that combines neural network encoders and probabilistic models to compress high-dimensional input and output spaces for efficient black-box optimization.
https://openreview.net/forum?id=wilej5VnqL
Compressor summary: InterLUDE is a new SSL method that improves image classification by interpolating labeled and unlabeled embeddings and minimizing discrepancies in predictions between them.
https://openreview.net/forum?id=weixEb6Wjd
Compressor summary: The paper examines how hardware choices can affect model performance and fairness in machine learning, highlighting the need for considering hardware in ML-as-a-service platforms.
https://openreview.net/forum?id=wea7nsJdMc
Compressor summary: The paper proposes a data structure for dynamic graph problems using algorithms-with-predictions to achieve consistency, robustness, and smoothness in running time, and shows empirical results on real datasets.
https://openreview.net/forum?id=wdezvnc9EG
Compressor summary: This paper investigates how different augmentations in graph contrastive learning affect downstream performance by separating classes rather than aligning nodes of the same class, and proposes two methods to verify the findings.
https://openreview.net/forum?id=wdTiuvd0fR
Compressor summary: The text discusses how large pretrained protein language models can improve protein prediction tasks but shows that their performance does not always scale with pretraining time and suggests a need for better pretraining methods.
https://openreview.net/forum?id=wWdkNkUY8k
Compressor summary: FastEGNN is an improved equivariant GNN model for large geometric graphs that uses virtual nodes to approximate the graph and achieve a balance between accuracy and efficiency.
https://openreview.net/forum?id=wUgTnf918v
Compressor summary: The paper proposes a spectral method to measure novelty in multi-modal generative models by comparing their sample types with a reference model using the Kernel-based Entropic Novelty score.
https://openreview.net/forum?id=wTd7dogTsB
Compressor summary: The paper analyzes how well score-based diffusion models can approximate complex distributions, and shows that they can achieve near-optimal performance under mild assumptions.
https://openreview.net/forum?id=wLoESsgZIq
Compressor summary: The paper introduces a method called local-DSM that allows training generative models with non-Gaussian priors using nonlinear diffusion processes and applies it to image generation and statistical physics problems.
https://openreview.net/forum?id=wK9RvVmi7u
Compressor summary: The paper proposes a general random graph model to analyze heterophily patterns in Graph Neural Networks (GNNs) and shows how different factors affect separability in GNNs.
https://openreview.net/forum?id=wGtzp4ZT1n
Compressor summary: Key points: - The paper explores competition dynamics in LLM-based agents using GPT-4 to simulate a virtual town with restaurant and customer agents - Competition encourages transformation, such as new operating strategies for restaurants - Simulation experiments reveal interesting findings that align with existing market and sociological theories Summary: The paper uses GPT-4 to create a competitive environment for LLM-based agents in a virtual town and studies how competition leads to transformation and insight into society.
https://openreview.net/forum?id=wG2SgnH6Zv
Compressor summary: The paper introduces a framework for learning unknown objects from training data using a given model class, with general applicability to various problems and learning guarantees based on the variation of the model class.
https://openreview.net/forum?id=wELbEYgnmo
Compressor summary: The paper argues for a more human-centered approach to AutoML, focusing on user interaction and collaboration between humans and AutoML systems.
https://openreview.net/forum?id=wDDprThYeT
Compressor summary: The paper introduces *xT*, a vision transformer framework that effectively aggregates global context with local details and can model large images without significant losses or memory growth.
https://openreview.net/forum?id=wDDGQabYPQ
Compressor summary: InferCept is a new framework for large language models that reduces resource waste and improves serving speed by efficiently handling interactions with external environments, tools, and agents.
https://openreview.net/forum?id=wCMNbdshcY
Compressor summary: Key points: - BEAST is a novel, fast, and interpretable adversarial attack for Language Models - BEAST can jailbreak aligned LMs quickly and with high success rates - BEAST can induce hallucinations in LM chatbots and generate privacy attacks Summary: BEAST is a new attack method that can break the security of Language Models, make them produce wrong or irrelevant outputs, and expose user information.
https://openreview.net/forum?id=wBr5ozDEKp
Compressor summary: The authors call for a better theoretical understanding of graph neural networks (GNNs), considering their expressive power, generalization, and optimization in graph machine learning.
https://openreview.net/forum?id=w9nxTXuaCc
Compressor summary: The paper proposes a new direction for machine learning research called $C^*$-algebraic ML, which unifies existing learning strategies and constructs a new framework using the mathematical concept of $C^*$-algebra.
https://openreview.net/forum?id=w8ei1o9U5y
Compressor summary: The paper introduces Multiagent Reinforcement Learning with Reset Replay (MARR), a method to improve sample efficiency of MARL in parallel environments using reset strategy and data augmentation.
https://openreview.net/forum?id=w8BnKGFIYN
Compressor summary: DART is a sample-efficient transformer-based method that uses discrete representations for world modeling and learning behavior, achieving high performance on Atari games without look-ahead search.
https://openreview.net/forum?id=w5oUo0LhO1
Compressor summary: Kernel SIVI (KSIVI) is a method for Bayesian inference that uses kernel tricks to simplify the score matching objective, enabling faster convergence and easier optimization.
https://openreview.net/forum?id=w4B42sxNq3
Compressor summary: ReeFL is a recurrent early exit approach for federated learning that fuses features from different sub-models into a single shared classifier, improving performance and privacy.
https://openreview.net/forum?id=w1d9DOGymR
Compressor summary: The paper explores how social choice theory can help address ethical and safety issues in fine-tuning foundation models like GPT-4 using human feedback or principles.
https://openreview.net/forum?id=w1HdBXSJXn
Compressor summary: The text introduces Persona In-Context Learning (PICLe), a method to customize the behavior of large language models based on desired personality traits using Bayesian inference and likelihood ratio selection criterion.
https://openreview.net/forum?id=vye4OgLaTy
Compressor summary: The paper presents a method called FlashST that improves traffic prediction by adapting pre-trained models to different scenarios and data distributions using a spatio-temporal prompt network and a distribution mapping mechanism.
https://openreview.net/forum?id=vxPmrxKe0J
Compressor summary: Probabilistic neurosymbolic models combine neural networks and logical reasoning, and WeightME is an unbiased gradient estimator for them based on model sampling.
https://openreview.net/forum?id=vxDjeeBnTu
Compressor summary: The paper analyzes two self-supervised learning methods using matrix mutual information and introduces M-MAE, a new method that improves upon existing ones for visual representation learning.
https://openreview.net/forum?id=vuMD71R20q
Compressor summary: The paper investigates how removing the square root from adaptive gradient optimizers improves generalization on convolutional architectures, while maintaining performance on transformers, and discusses practical benefits for developing non-diagonal adaptive methods.
https://openreview.net/forum?id=vsy21Xodrt
Compressor summary: GraphECL is a fast and efficient contrastive learning method for graphs that uses an MLP to mimic the computations of a GNN, achieving superior performance and efficiency compared to existing methods.
https://openreview.net/forum?id=vsOF7qDNhl
Compressor summary: This paper studies how stochastic gradient descent behaves in non-convex problems and shows that it visits low-energy regions, like the global minimum, more often than high-energy ones, using a thermodynamics analogy.
https://openreview.net/forum?id=vq7ITv8a49
Compressor summary: Kernel debiased plug-in estimation is a novel method to estimate multiple target parameters in nonparametric models without relying on influence functions and while maintaining computational efficiency.
https://openreview.net/forum?id=vn92qYjL1F
Compressor summary: The paper proposes a new transductive algorithm that improves robustness against adversarial attacks by applying a reduction technique to construct effective defenses with provable guarantees.
https://openreview.net/forum?id=vl9GB3fbht
Compressor summary: The authors propose a method for learning differential and integral operators with locally supported kernels to improve the performance of Fourier neural operators for solving partial differential equations.
https://openreview.net/forum?id=vjkq5fwsj3
Compressor summary: The paper proposes a method to construct neural networks that can learn from data on graphs with complex relations using their actual symmetry group, Aut(G), instead of the symmetric group S_n.
https://openreview.net/forum?id=vhMq3eAB34
Compressor summary: DiffDA is a denoising diffusion model that uses GraphCast and predicted states to generate high-resolution atmospheric data from sparse observations, enabling weather forecasting and climate modeling applications.
https://openreview.net/forum?id=veEjiN2w9F
Compressor summary: The authors propose a framework using computational complexity theory to assess local and global perspectives of interpreting ML models, comparing linear models, decision trees, and neural networks.
https://openreview.net/forum?id=va3r3hSA6n
Compressor summary: The paper investigates the relationship between absolute and relative positional encodings for graph transformers, finding them equivalent in distinguishing non-isomorphic graphs, but with potential differences depending on node features.
https://openreview.net/forum?id=vYYIuJDTHq
Compressor summary: The paper presents partial solutions to the linear ordering problem by constructing maps and testing cost function conditions.
https://openreview.net/forum?id=vXUqOCsbj8
Compressor summary: We analyze the efficiency of modern Hopfield models based on pattern norms and show sub-quadratic variants exist under a computational complexity assumption, with examples and lower bounds provided.
https://openreview.net/forum?id=vSerUPYFtB
Compressor summary: The text proposes a universal perturbation generator that uses multi-modal pre-trained models to transform image data into text concepts and make it unlearnable, protecting personal information in free internet data.
https://openreview.net/forum?id=vQmVmMN5ft
Compressor summary: The paper proposes a new space for gradient compression in federated learning, which improves compressibility and allows for higher accuracies with minimal information loss.
https://openreview.net/forum?id=vMUnnS4OWC
Compressor summary: The paper proposes a new method to sample from unnormalized probability distributions using a particle-based denoising diffusion scheme with a novel score matching loss, which is more consistent than standard methods.
https://openreview.net/forum?id=vLtVGtEz5h
Compressor summary: The paper proposes a fast and parallelizable distance for comparing spherical probability distributions using the stereographic projection and generalized Radon transform, called Stereographic Spherical Sliced Wasserstein (S3W) distance, and evaluates its speed and accuracy in various applications.
https://openreview.net/forum?id=vKtomqlSxm
Compressor summary: Chain of Code is an extension that improves language models' ability to reason by having them write and emulate code for semantic tasks.
https://openreview.net/forum?id=vJx6fld6l0
Compressor summary: The study presents HEPT, a novel efficient transformer model for large-scale point cloud processing in HEP and astrophysics, using local inductive bias and OR & AND-construction LSH for kernel approximation.
https://openreview.net/forum?id=vITl6CqIkk
Compressor summary: The paper proposes a framework to improve text-to-image generative models for human-object interaction by refining human pose and interaction boundary regions using guidance from pose quality and boundary information.
https://openreview.net/forum?id=vGHOFeUQi8
Compressor summary: NQE is a new SBI method that uses conditional quantile regression and spline interpolation to estimate posterior samples and credible regions, with calibration options for limited data or model errors.
https://openreview.net/forum?id=vG7YpsJT74
Compressor summary: The paper proposes a new method, GraCe, for nonconvex ZOO in high dimensions, which uses fewer queries and achieves similar convergence rates as previous methods.
https://openreview.net/forum?id=vFk9fqXLst
Compressor summary: The paper highlights the importance of considering equivariant models' inductive bias when using latent representations and proposes principles for choosing invariant projections to improve performance in downstream tasks, as shown in molecular graph generation and image classification examples.
https://openreview.net/forum?id=vFATIZXlCm
Compressor summary: The paper presents a mathematically rigorous neural solver for linear PDEs that combines U-Net and MultiGrid principles, achieves high accuracy and generalization, and introduces a new residual loss metric for unsupervised training.
https://openreview.net/forum?id=vCN5lwcWWE
Compressor summary: The paper proposes Lookbehind, a method that improves the loss-sharpness trade-off in SAM by using multiple ascent steps and linear interpolation, leading to better generalization, robustness, and lifelong learning performance.
https://openreview.net/forum?id=vBJZ93tvoE
Compressor summary: The authors use graph neural networks (GNNs) to model bacterial communities from their genomes, predicting relative abundance profiles without growth curves, and show generalization to unseen bacteria and different community structures.
https://openreview.net/forum?id=v9tIJW1fzt
Compressor summary: The authors propose using low-precision floating-point representations to reduce energy consumption in Gaussian process regression, and show that well-conditioned kernel matrices allow for significant energy savings without compromising model performance.
https://openreview.net/forum?id=v8MgLJ7kbL
Compressor summary: The paper proposes an adaptive rolling window method for model assessment and selection that works well with changing environments and historical data.
https://openreview.net/forum?id=v7I5FtL2pV
Compressor summary: Charms is a method to transfer relevant tabular knowledge to images, enhancing image classification and interpretability.
https://openreview.net/forum?id=v6tAdeCXKH
Compressor summary: The paper proposes FedSaC, a framework for Federated Learning that balances similarity and complementarity among clients to enhance model performance in heterogeneous and multimodal scenarios.
https://openreview.net/forum?id=v6eaD7Wekw
Compressor summary: The paper proposes an adaptive method for learning from corrupted data, using latent variables and variational inference to infer the corruption level and improve performance on various machine learning tasks.
https://openreview.net/forum?id=v2o9rRJcEv
Compressor summary: The paper proposes a new method, TDPO-R, that aligns diffusion models with human preferences while minimizing reward overoptimization by exploiting the temporal inductive bias of diffusion models and addressing primacy bias from active neurons in the critic model.
https://openreview.net/forum?id=v1I4zRAjMb
Compressor summary: The paper presents TENG, a novel method for solving time-dependent PDEs using neural networks with high accuracy and efficiency.
https://openreview.net/forum?id=v0VUsQI5yw
Compressor summary: Smoothness Adaptive Transfer Learning (SATL) is a kernel ridge regression algorithm that adapts to varying and unknown smoothness in transfer learning tasks, achieving minimax optimality and optimal statistical rate.
https://openreview.net/forum?id=uyhjKoaIQa
Compressor summary: The paper proposes a new method to improve GNNs by constructing a graph from features using Delaunay Triangulation, which reduces over-squashing and oversmoothing problems.
https://openreview.net/forum?id=uydQ2W41KO
Compressor summary: RLAIF is an alternative to RLHF that uses preferences from an LLM and achieves comparable or better results than RLHF, overcoming scalability issues.
https://openreview.net/forum?id=uun4fzaiat
Compressor summary: The paper proposes a greedy method for selecting samples with large approximate losses instead of exact losses to reduce selection overhead and training time, and evaluates it on BERT model training.
https://openreview.net/forum?id=uubBZKM99Y
Compressor summary: Flora reduces memory usage for training large neural networks by using random projections and resampling matrices, without sacrificing performance.
https://openreview.net/forum?id=usUPvQH3XK
Compressor summary: The text proposes a framework that combines large language models with reinforcement learning to create strategic language agents that can overcome intrinsic bias and perform well in complex decision-making tasks like the Werewolf game.
https://openreview.net/forum?id=us6zMORsMe
Compressor summary: The study develops a novel method (MRM-GP) that combines Linear Dynamical System and Gaussian Process techniques to analyze complex interactions between different brain regions, revealing communication patterns and separating them by frequency bands.
https://openreview.net/forum?id=uqWfZ23O9g
Compressor summary: AMORE is a new method for learning laws in hybrid dynamical systems that jointly discovers equations and categorizes modes, outperforming existing two-stage approaches.
https://openreview.net/forum?id=upO8FUwf92
Compressor summary: Compositional Concept Extraction (CCE) is a new method to find high-level concepts in foundation models that are useful for understanding them, and it outperforms existing methods on various datasets and tasks.
https://openreview.net/forum?id=uog14iBFLA
Compressor summary: The paper proposes a new algorithm for multi-task learning in linear contextual bandits that improves efficiency by using representation learning and provides theoretical guarantees on sample and iteration complexity.
https://openreview.net/forum?id=uku9r6RROl
Compressor summary: This paper explores how level sampling affects generalization in deep RL agents, proposes data-regularised environment design to improve it, and provides theoretical justification for prioritizing levels based on value loss.
https://openreview.net/forum?id=uiqbnV4msl
Compressor summary: The paper explores how different aspects of neural ODE training impact performance and introduces a new initialization technique based on stability analysis.
https://openreview.net/forum?id=ui8ewXg1hV
Compressor summary: The paper proposes and analyzes dual propagation, a local learning algorithm that mimics gradients by using oppositely nudged compartments in artificial neurons, and shows its relation to adversarial robustness and a stable adjoint state method.
https://openreview.net/forum?id=uhHDhVKFMW
Compressor summary: The paper proposes LESS, a cache method that integrates a constant-sized cache with eviction-based methods to improve memory efficiency and retain information in large language models.
https://openreview.net/forum?id=ugxGpOEkox
Compressor summary: The authors study how safety prompts affect large language models' behavior and propose a method to optimize them for better safeguarding against harmful queries.
https://openreview.net/forum?id=ug2uoAZ9c2
Compressor summary: SANE is a method to learn task-agnostic representations of neural networks that can embed larger models into a learned space and generate unseen models sequentially.
https://openreview.net/forum?id=ufgVvFmUom
Compressor summary: The text introduces a sparse Johnson-Lindenstrauss transform that can embed points into fewer dimensions while preserving distances, and improves upon previous sparsity results for certain cases.
https://openreview.net/forum?id=ufCptn28vG
Compressor summary: Isometric Diffusion is a technique that improves diffusion models by learning a disentangled latent space, enabling smoother interpolation, more accurate inversion, and better control over attributes.
https://openreview.net/forum?id=udFZhUgtkI
Compressor summary: Key points: - SASNet is a novel method to restore blurred images with event streams at arbitrary scales - It uses spatial and temporal correlation features from events to generalize at continuous scales - It has two modules: SIRM for local areas and TIRM for global motion blur - A new H2D dataset is introduced to evaluate the method Summary: SASNet restores blurred images with event streams using spatial and temporal correlation features from events, and has two modules for different scales of blur. It is tested on a new dataset.
https://openreview.net/forum?id=ud4GSrqUKI
Compressor summary: The paper explores how to identify and measure the uncertainty in large language models' outputs using small probes and unsupervised methods.
https://openreview.net/forum?id=ucl3B05EsX
Compressor summary: PHAZE optimizes deep learning training by co-optimizing architecture and device placement using novel algorithms, achieving higher throughput than TPUv4 and Spotlight on large language models.
https://openreview.net/forum?id=uaExqhJ2Ag
Compressor summary: This paper presents a near-optimal solution for sparse recovery and other tasks in white-box adversarial streaming models without needing a random oracle, using homomorphic encryption schemes.
https://openreview.net/forum?id=uYISs2tpwP
Compressor summary: The text proposes a framework called conformal factuality that ensures high probability correctness guarantees for language models by connecting language modeling and conformal prediction, which involves making LM outputs less specific to expand uncertainty sets.
https://openreview.net/forum?id=uYIFQOtb58
Compressor summary: The paper proposes a method for spatiotemporal forecasting that handles missing data by coarsening time series and combining them with attention.
https://openreview.net/forum?id=uWNUTRgBso
Compressor summary: The text introduces new generalization bounds for machine learning algorithms using slicing methods, which improve performance and enable control over compressibility in high-dimensional problems.
https://openreview.net/forum?id=uUeXaKLE1I
Compressor summary: The paper studies the submodular cover problem with updates to the set, proposing a randomized algorithm that approximates the optimal solution and minimizes queries per update.
https://openreview.net/forum?id=uTC9AFXIhg
Compressor summary: The authors propose a unified framework for developing, integrating, and improving LLM-based agents using computational graphs and automatic graph optimizers.
https://openreview.net/forum?id=uRz9GZN17X
Compressor summary: The paper proposes a method for few-shot semantic segmentation that uses feedback and rectification layers to address the impact of intra-class diversity and improve the model's performance.
https://openreview.net/forum?id=uQiFsBil3p
Compressor summary: The authors propose a random matrix theory method for estimating Fréchet means on symmetric positive definite matrices in machine learning tasks, which performs better than existing methods with low sample support and many matrices to average.
https://openreview.net/forum?id=uQ2FUoFjnF
Compressor summary: LLMCompiler is a tool that improves the efficiency, cost, and accuracy of large language models by enabling parallel function calling for complex tasks.
https://openreview.net/forum?id=uN39Tt9P8b
Compressor summary: The paper proposes a new sequential method for predicting regions of multivariate time series with good coverage and small regions.
https://openreview.net/forum?id=uLpyWQPyF9
Compressor summary: ES-MoE is a new method that improves the efficiency of MoE training by balancing token loads and using host memory, achieving better scalability and throughput than existing frameworks.
https://openreview.net/forum?id=uLonuOfrwp
Compressor summary: The study proposes a statistical test for ViT's attention mechanisms in computer vision tasks, enabling reliable quantitative evidence for decision-making with controlled error rates, applied to brain image diagnosis.
https://openreview.net/forum?id=uKmcyyrZae
Compressor summary: The paper questions the benefits of global attention in Graph Transformer for graph-structured data, proposes a Bi-Level Global Graph Transformer with Collaborative Training to address the over-globalizing problem, and provides empirical and theoretical evidence of its effectiveness.
https://openreview.net/forum?id=uGoi3nY62g
Compressor summary: The paper introduces a new black-box attack method to exploit vulnerabilities in pixel-wise regression models for applications like autonomous driving and augmented reality, showing its effectiveness against 7 models and Google's online service.
https://openreview.net/forum?id=uEx2bSAJu8
Compressor summary: The paper proposes a graph-based multi-view clustering algorithm that infers the unknown number of clusters $K$ using inter-cluster connectivity as a reward in reinforcement learning, outperforming existing methods.
https://openreview.net/forum?id=uDoy7AGvEC
Compressor summary: LayerMerge is a novel depth compression method that jointly prunes convolution layers and activation functions to enhance efficiency without sacrificing performance.
https://openreview.net/forum?id=uDkXoZMzBv
Compressor summary: This paper proposes a method to reduce computational costs of overparameterized models by exploiting low-dimensional structures in data and model parameters, showing its effectiveness for deep matrix completion and language model fine-tuning.
https://openreview.net/forum?id=uCdcXRuHnC
Compressor summary: The authors propose a machine learning method using neural density estimation to efficiently infer population parameters in non-linear mixed-effects models for heterogeneous populations in various fields.
https://openreview.net/forum?id=uA3FRvO2DJ
Compressor summary: The paper introduces a new framework to better understand and improve normalizing flows, which are neural networks for transforming one probability distribution into another, by showing their limitations and how to overcome them.
https://openreview.net/forum?id=u9qmjV2khT
Compressor summary: The paper characterizes the properties of local and global optima of the MCR$^2$ objective for learning structured deep representations, showing that it leads to diverse and discriminative solutions.
https://openreview.net/forum?id=u9oSQtujCF
Compressor summary: Key points: - Graph neural networks need to generalize out-of-distribution data - Existing methods assume spurious features and target labels are correlated - New paradigm induces a robust inductive bias based on infomax principle - EQuAD framework disentangles invariant and spurious features - EQuAD shows improved performance in synthetic and real datasets Summary: The paper proposes a novel graph invariance learning method, EQuAD, that leverages the infomax principle to learn invariant features and disentangle them from spurious ones, achieving better generalization on OOD data.
https://openreview.net/forum?id=u8TZ9gm4im
Compressor summary: The text describes a new image compression algorithm that uses text information to improve both perceptual and pixel-wise quality, while avoiding the drawbacks of existing methods.
https://openreview.net/forum?id=u6PeRHEsjL
Compressor summary: The text suggests that large language models can shape each other's behavior and form emergent AI societies, which can have benefits for human society and online environments, and calls for research on these issues.
https://openreview.net/forum?id=u4VR3WBH7a
Compressor summary: Key points: - The paper proposes a new method for continual audio-video pre-training that addresses two challenges: sparse spatio-temporal correlation and multimodal correlation overwriting. - The method uses Localized Patch Importance Scoring and Replay-guided Correlation Assessment to select relevant patches for pre-training. - The method improves performance in zero-shot retrieval tasks and reduces memory consumption. Summary: The paper presents a new continual audio-video pre-training method that uses patch importance scoring and correlation assessment to select relevant patches, improving retrieval performance and memory efficiency.
https://openreview.net/forum?id=u26c52rxZC
Compressor summary: The study uses patent data to improve antibody humanness prediction by training an encoder with contrastive learning and cross-entropy loss.
https://openreview.net/forum?id=u09gadH3BU
Compressor summary: The paper introduces any-precision LLM, a method to compress large language models and deploy them efficiently using varying bit-widths without sacrificing quality or performance.
https://openreview.net/forum?id=u00dmbI8Db
Compressor summary: The MFAS framework improves egocentric video question answering by enhancing small object recognition, suppressing noise, and aggregating visual semantics guided by questions.
https://openreview.net/forum?id=tya725xlZ3
Compressor summary: The paper proposes a deep network that learns generative-to-discriminative representations for masked face recognition, using synthetic masked faces for pretraining.
https://openreview.net/forum?id=txRZBD8tBV
Compressor summary: The paper studies the asymmetry in importance of low-rank adapter matrices during fine-tuning and shows that updating B matrix is more effective than A matrix, leading to parameter savings and improved generalization.
https://openreview.net/forum?id=twm7qPVX1F
Compressor summary: The paper proposes a new Bayesian method to identify causal directions in Markov equivalent structures, which works better than maximum likelihood methods and can handle realistic assumptions.
https://openreview.net/forum?id=tw1PwpuAuN
Compressor summary: The authors propose a novel method for measuring the faithfulness of NLP model explanations by incorporating token masking into a fine-tuning process that makes it in-distribution and improves importance measures.
https://openreview.net/forum?id=tu5fCCuua2
Compressor summary: This paper studies how the amount of computation time and memory affects the performance of a neural solver called Differentiable Neural Computer (DNC) on various algorithms, and finds that limiting its planning steps can lead to poor generalization and stability.
https://openreview.net/forum?id=ttnbM598vZ
Compressor summary: Pairwise Alignment (Pair-Align) is a novel graph domain adaptation method that handles shifts in features, labels, and connecting patterns using edge weights and label weights to recalibrate node influence and adjust classification loss.
https://openreview.net/forum?id=ttaTyweIr1
Compressor summary: The paper presents a neurosymbolic system that learns to infer general-purpose programs from visual datasets to perform various tasks in different domains.
https://openreview.net/forum?id=tpYHbEl7P1
Compressor summary: The text discusses how to define and find flat minima in optimization algorithms for machine learning applications using the trace of the Hessian as a measure of flatness.
https://openreview.net/forum?id=tp6ruPIfIV
Compressor summary: The paper shows that posterior sampling for inpainting and other tasks is computationally intractable even with fast unconditional sampling, using cryptographic assumptions.
https://openreview.net/forum?id=tmUorldOWN
Compressor summary: This paper explores a new security vulnerability in machine unlearning methods, proposes an attack that reduces adversarial robustness, and shows its potential for enhancing model stealing attacks.
https://openreview.net/forum?id=tl2qmO5kpD
Compressor summary: Offline actor-critic reinforcement learning can scale to large models like transformers and outperform behavioral cloning for multi-task training on continuous control tasks with sub-optimal data.
https://openreview.net/forum?id=tgsSKziIEa
Compressor summary: KPOD is a framework that improves language model reasoning transfer by weighting tokens and using progressive distillation.
https://openreview.net/forum?id=teteOa9nJ9
Compressor summary: BSP-WSA is a novel method for universal domain adaptation that uses an adversarial classifier, singular value decomposition, and weighted semantic augmentation to handle category shift between domains.
https://openreview.net/forum?id=teHPKqjX8q
Compressor summary: The paper proposes a diagnosis method for neural networks based on loss landscape metrics, which outperforms validation-based approaches in identifying the source of model failure.
https://openreview.net/forum?id=tdomF3PW6A
Compressor summary: The paper proposes Convex Concave Loss (CCL), a method to increase the loss variance of training data and defend against membership inference attacks by reducing the convexity of loss functions with a concave term.
https://openreview.net/forum?id=tc3Nmcpmnx
Compressor summary: The text discusses how to improve sample-based inference for Bayesian neural networks by understanding the relationship between weight and function space, and provides guidelines and an effective solution with competitive performance and uncertainty quantification.
https://openreview.net/forum?id=tVwzR1myUp
Compressor summary: The Continuum Physical Dataset (ContPhy) is a new benchmark for testing AI models' ability to reason about diverse physical properties and dynamics of soft-bodied objects and other continuum phenomena, revealing their current limitations and inspiring improvements in perception and reasoning.
https://openreview.net/forum?id=tTtSnpH4fc
Compressor summary: The paper proposes a new algorithm, PIECE, that achieves finite-time logarithmic regret bounds for self-tuning regulation problem and improves initial transient performance.
https://openreview.net/forum?id=tTq3qMkJ8w
Compressor summary: The paper proposes CooK, a model that uses co-occurrence knowledge and TF-$l$-IDF to improve scene graph generation, achieving better performance and generalization than existing methods.
https://openreview.net/forum?id=tSjyKR8WIf
Compressor summary: The paper proposes that neural noise can help humans and AI models group and segment images without supervision, leading to better performance on perceptual grouping tasks.
https://openreview.net/forum?id=tRESfzWFtf
Compressor summary: The paper presents new interior-point methods for non-convex optimization with constraints that have better global complexity than existing methods.
https://openreview.net/forum?id=tQPkzTdaaN
Compressor summary: PARDEN is a method that uses a large language model as a safeguard to detect and prevent jailbreaks by asking it to repeat its own outputs, achieving better results than existing approaches.
https://openreview.net/forum?id=tOO6PD3kYP
Compressor summary: The paper proposes and analyzes a Bayesian optimization method using Gaussian Process models with random exploration, which achieves optimal error rates, has computational advantages, and partially resolves an open problem.
https://openreview.net/forum?id=tMkPL7Tiul
Compressor summary: The text presents a new method for creating sketches that can be applied to rank one tensors in two or three modes, achieving faster computation times by using fast convolution and random subsets of tensor entries.
https://openreview.net/forum?id=tHBLwSYnLf
Compressor summary: The paper proposes the function encoder, an algorithm that helps reinforcement learning agents transfer between related tasks by providing a coherent vector representation of the reward or transition function.
https://openreview.net/forum?id=tFEOOH9eH0
Compressor summary: The text introduces a new dataset of visiontouch pairs with English labels and presents a TVL model that improves tactile-vision-language alignment over existing models.
https://openreview.net/forum?id=tDRYrAkOB7
Compressor summary: Dynamic Memory Compression (DMC) is a method to compress the cache of key-value representations for past tokens in large language models, improving generation efficiency and allowing them to fit longer contexts and larger batches within any given memory budget.
https://openreview.net/forum?id=tDMlQkJRhZ
Compressor summary: SPHINX-X is a large language model developed upon SPHINX, with improved architecture and training efficiency, trained on diverse multi-domain and multi-modal data, and obtaining strong correlation between performance and data/parameter scales.
https://openreview.net/forum?id=tASXcrMekp
Compressor summary: MADA is a unified optimizer framework that learns the best adaptive optimizer for deep learning tasks during training, and outperforms Adam and other popular optimizers on vision and language tasks.
https://openreview.net/forum?id=tABvuya05B
Compressor summary: The paper proposes a method for continuous learning that partitions the network into three parts to share knowledge between old and new tasks, improving performance and preventing forgetting.
https://openreview.net/forum?id=t8mt4YrPsq
Compressor summary: Larimar is a brain-inspired architecture that enhances LLMs with a distributed episodic memory for efficient, accurate, and flexible knowledge updates without re-training or fine-tuning.
https://openreview.net/forum?id=t8WDBcegae
Compressor summary: The paper shows that finding approximate second-order stationary points in non-convex optimization problems is as hard as finding first-order stationary points and depends on the domain complexity, contrary to previous results.
https://openreview.net/forum?id=t82Y3fmRtk
Compressor summary: The paper presents **R**$^3$, a method that uses outcome supervision and reverse curriculum to improve large language models' reasoning skills.
https://openreview.net/forum?id=t6dBpwkbea
Compressor summary: The paper proposes TimeX++, an explanation framework that uses information bottleneck principle to produce high-quality explanations for deep learning models operating on time series data, and evaluates it on synthetic and real-world datasets.
https://openreview.net/forum?id=t4908PyZxs
Compressor summary: The paper proposes a cognitive-inspired method for few-shot class-incremental learning that uses compositional learning, set similarities, and primitive reusability to recognize novel classes with fewer samples, achieving improved performance and interpretability.
https://openreview.net/forum?id=t3SEfoTaYQ
Compressor summary: Coprocessor Actor Critic is a novel reinforcement learning method for adaptive brain stimulation that learns how to act and induce optimal actions in the world with less samples and higher success than traditional approaches.
https://openreview.net/forum?id=szxtVHOh0C
Compressor summary: Key points: - Surface-VQMAE is a novel surface-based unsupervised learning algorithm for protein function analysis - It uses Transformer architecture and vector quantization to capture patch-level relations and enforce discrete posterior distribution - It shows effectiveness in various scenarios such as binding site scoring, affinity prediction, and mutant effect estimation Summary: Surface-VQMAE is a new algorithm that uses protein surface information and unsupervised learning to analyze protein functions, achieving good results in different applications.
https://openreview.net/forum?id=szvKJgmubh
Compressor summary: Key points: - The text is about achieving adversarial robustness without real data - Existing methods assume accessibility to original data and fail in this scenario - DataFreeShield proposes two solutions: surrogate dataset generation and adversarial training - DataFreeShield outperforms baselines and shows the first data-free solution Summary: The text presents DataFreeShield, a method that achieves adversarial robustness without real data by generating surrogate datasets and training with them, outperforming existing methods.
https://openreview.net/forum?id=szRHR9XGrY
Compressor summary: The paper introduces BiDST, a novel framework for dynamic sparse training that optimizes both weights and masks simultaneously, achieving better accuracy, faster speed, and reduced overhead compared to traditional methods.
https://openreview.net/forum?id=syXFAVqx85
Compressor summary: Dirichlet flow matching on the simplex improves sequence generation speed and quality over autoregressive models, especially for complex DNA sequences.
https://openreview.net/forum?id=swTG6xju8O
Compressor summary: The paper proposes IM-3D, a text-to-3D model that uses video generators, 3D reconstruction with Gaussian splatting, and reduces evaluation times to create high-quality 3D outputs efficiently.
https://openreview.net/forum?id=svm53KQAtN
Compressor summary: EOI is a novel sparse orthogonal initialization scheme that provides exact orthogonality and enables creation of layers with arbitrary densities, outperforming common sparse initialization techniques for static sparse training.
https://openreview.net/forum?id=stMhi1Sn2G
Compressor summary: Accelerated Speculative Sampling (ASpS) is a new algorithm that improves inference speed of large language models by using Tree Monte Carlo methods to generate multiple tokens in one step and finding better global maximum couplings.
https://openreview.net/forum?id=st2BTty53v
Compressor summary: The paper proposes a new method for protecting anonymity in facial recognition by creating adversarial obfuscation that fools face restoration techniques, which can restore pixelated faces with high accuracy.
https://openreview.net/forum?id=ssFMq35UUY
Compressor summary: The paper proposes ULAREF, a unified framework for learning with inaccurate supervision, which refines labels using global reliability detection and local enhancement with a consistency loss.
https://openreview.net/forum?id=srejp9uOx7
Compressor summary: The authors propose using ReLU neurons with constraints in deep neural networks to learn implicit neural representations, showing their effectiveness and versatility in various tasks.
https://openreview.net/forum?id=sqv2xP8rfb
Compressor summary: The paper proposes a new method for abductive learning, called Ambiguity-Aware Abductive Learning (A$^3$BL), which improves the existing approach by evaluating all potential candidates and their probabilities to better handle uncertainty in the knowledge base.
https://openreview.net/forum?id=spOpHW1No2
Compressor summary: This paper proves that the worst case teaching dimension occurs when using binary representations of numbers in machine teaching.
https://openreview.net/forum?id=snhurpZt63
Compressor summary: The paper introduces a new underwater salient instance segmentation dataset (USIS10K) and a method (USIS-SAM) based on Segment Anything Model that uses visual prompts to improve segmentation accuracy in underwater scenes.
https://openreview.net/forum?id=sjJZHPV9Id
Compressor summary: The paper proposes AEMatter, a new matting network with a Hybrid-Transformer backbone and appearance-enhanced axis-wise learning blocks that achieves superior performance compared to existing methods.
https://openreview.net/forum?id=shzEkKPrsn
Compressor summary: The paper proposes a new online learning framework that works without strict feasibility assumptions and proves its no-regret guarantees for bidding in ad auctions.
https://openreview.net/forum?id=sfQH4JJ4We
Compressor summary: The paper presents fast and scalable algorithms for hypergraph models in semi-supervised learning that capture higher-order relations better than graph-based methods and improve clustering on categorical data.
https://openreview.net/forum?id=seo9V9QRZp
Compressor summary: Gradual magnitude pruning improves value-based deep reinforcement learning agents' parameter effectiveness and performance with minimal parameters.
https://openreview.net/forum?id=scSB9RynSd
Compressor summary: The text discusses how machine learning models' performance improves with more data and how individual data points contribute differently to model accuracy depending on dataset size, suggesting ways to use this knowledge for data valuation and selection.
https://openreview.net/forum?id=scMAQ3mFAA
Compressor summary: Cost-aware node selection improves Bayesian optimization of function networks by reducing evaluation costs.
https://openreview.net/forum?id=scFlbJQdm1
Compressor summary: The authors propose a novel framework that uses a transformer-based model to generate new chemical structures while ensuring synthetic accessibility through postfix notations of synthesis pathways.
https://openreview.net/forum?id=sb81Xl50JG
Compressor summary: APT is a method that adaptively prunes and tunes language model parameters for improved training and inference efficiency, maintaining high performance with significantly reduced parameters and faster fine-tuning.
https://openreview.net/forum?id=saP7s0ZgYE
Compressor summary: The paper proposes a fast and efficient algorithm for correlation clustering in dynamic graphs, with performance guarantees similar to the well-known Pivot algorithm.
https://openreview.net/forum?id=sZla6SnooP
Compressor summary: The paper proposes model-based policy iteration algorithms for nonlinear optimal control problems using neural networks to solve PDEs, with convergence guarantees and outperforming traditional methods.
https://openreview.net/forum?id=sYeioWoF9u
Compressor summary: The paper proposes LMIndexer, a self-supervised framework that uses a generative language model to learn semantic identifiers for information retrieval tasks, addressing challenges such as sequential discrete ID and semantic supervision deficiency.
https://openreview.net/forum?id=sTVSyqD6XX
Compressor summary: The paper proposes methods for preserving privacy in Federated Learning with centralized systems, ensuring Differential Privacy while maintaining optimal convergence rates and linear computational complexity.
https://openreview.net/forum?id=sT7UJh5CTc
Compressor summary: The paper proposes a new statistical test for robustly detecting if a data point was used to train a model, using reference models and population data samples, with low computational cost and high accuracy.
https://openreview.net/forum?id=sSAEhcdB9N
Compressor summary: This paper proposes novel federated learning algorithms that protect data privacy, perform well with heterogeneous data, and require fewer communication rounds than existing methods.
https://openreview.net/forum?id=sOyJSNUrzQ
Compressor summary: The paper presents a generic framework called ReED for analyzing various knowledge graph representation learning (KGRL) models and provides theoretical generalization bounds for them, which can guide their practical design.
https://openreview.net/forum?id=sO5qtpvsUZ
Compressor summary: The Optimal Eye Surgeon (OES) framework prunes and trains deep image generator networks by adaptively underparameterizing them, which helps resist overfitting to noise and improves image restoration tasks.
https://openreview.net/forum?id=sNjxqSnXFO
Compressor summary: The paper presents quantum algorithms for sampling from non-logconcave probability distributions and estimating their partition functions, overcoming challenges by using a reference reversible Markov chain and showing polynomial speedups.
https://openreview.net/forum?id=sLZzFTMWSt
Compressor summary: CaRiNG is a method to identify causal processes in sequential data with non-invertible generation, using temporal context and an identifiability theory.
https://openreview.net/forum?id=sKjcrAC4eZ
Compressor summary: The paper studies how Lie group invariances reduce sample complexity and improve convergence rates when estimating various distances and density estimation problems in machine learning models.
https://openreview.net/forum?id=sHtIStlg0v
Compressor summary: The authors study geographic biases in large language models and show that they are correlated with socioeconomic conditions and exhibit significant variation across models.
https://openreview.net/forum?id=sHswzNWUW2
Compressor summary: The paper proposes a new language-driven framework for zero-shot multi-label image recognition using CLIP, LLMs, and cross-modal mapping without annotated images during training.
https://openreview.net/forum?id=sFN49CfklF
Compressor summary: LaGTran is a text supervision framework that transfers discriminative knowledge from labeled source to unlabeled target data with domain gaps, outperforming prior approaches on image and video datasets.
https://openreview.net/forum?id=sF9epWkNUG
Compressor summary: VCNeFs are a new method to solve time-dependent PDEs using neural fields, attention mechanisms, and conditional learning, addressing limitations of previous Transformer-based approaches.
https://openreview.net/forum?id=sDjszMb2Ir
Compressor summary: LASER is a new compression scheme that efficiently transmits low-rank gradients over noisy channels, achieving better performance than existing methods on practical machine learning tasks.
https://openreview.net/forum?id=sCGRhnuMUJ
Compressor summary: The paper proposes JSQ, a novel model compression technique for large language models that integrates sparsification and quantization to achieve significant computation reduction without performance degradation.
https://openreview.net/forum?id=sBJNokmYuV
Compressor summary: CPL is a method that generates refined candidate pseudolabels for vision-language models to improve fine-tuning performance in downstream tasks.
https://openreview.net/forum?id=s9RKqT7jVM
Compressor summary: The text discusses concerns about the stability of deep neural policies in various settings, and introduces a method to analyze and improve these policies using robust training techniques.
https://openreview.net/forum?id=s7RDnNUJy6
Compressor summary: Weight Averaged Reward Models (WARM) is a method to prevent reward hacking in large language models by averaging multiple fine-tuned reward models, improving quality and alignment of predictions.
https://openreview.net/forum?id=s6ZAT8MLKU
Compressor summary: The paper analyzes the convergence rates of two variants of the GDA algorithm for minimax optimization problems, shows that Alternating-GDA is faster and has better performance than Simultaneous-GDA, and proposes a new algorithm called Alex-GDA that improves upon both.
https://openreview.net/forum?id=s5PLISyNyP
Compressor summary: The paper proposes a method to estimate the CATE from observational data in multiple environments with violations of standard causal assumptions by using them as instrumental variables and combining meta-learners with machine learning models.
https://openreview.net/forum?id=s4h6nyjM9H
Compressor summary: T-RevSNN is a novel spiking neural network architecture that reduces memory, training time, and inference energy costs by temporally reversible interactions and redesigned input encoding.
https://openreview.net/forum?id=s4Hy0L4mml
Compressor summary: The paper proposes a novel approach (MS-TIP) that uses transformers, multi-scale hypergraphs, and scenic attention to predict pedestrian trajectories accurately even when the observed sequences are incomplete.
https://openreview.net/forum?id=s4EYBJ30WY
Compressor summary: The paper introduces spectral imbalance as a source of class bias in feature learning and proposes a framework to study, compare, and mitigate it using 11 pre-trained encoders.
https://openreview.net/forum?id=s3e8poX3kb
Compressor summary: The authors study why large language models make factual errors, find a pattern related to sharpness of context activations, and propose an entropy-based metric to improve the quality of generated text by adjusting token prediction distributions.
https://openreview.net/forum?id=s1sdx6vNsU
Compressor summary: This paper analyzes how low-rank adaptation (LoRA) helps fine-tune large language models efficiently and effectively by avoiding spurious minima and achieving good generalization.
https://openreview.net/forum?id=s0UDX7Kswl
Compressor summary: DiffAug is a novel unsupervised contrastive learning technique that uses diffusion models to generate positive samples based on semantic encoders, improving representation ability in various domains.
https://openreview.net/forum?id=s0Jvdolv2I
Compressor summary: Feature visualizations are not reliable for explaining how neural networks process natural images because they can be easily fooled by arbitrary patterns and do not match standard input processing.
https://openreview.net/forum?id=ryDa4mS18V
Compressor summary: SAM-E is a novel architecture for robot manipulation that uses a vision foundation model for scene understanding and sequence imitation for long-term action reasoning, achieving superior performance and efficiency.
https://openreview.net/forum?id=rvaN2P1rvC
Compressor summary: DAIS optimizes a symmetrized KL divergence between initial and target distribution, acting as a variational inference method that often improves uncertainty estimates.
https://openreview.net/forum?id=rucbIsWoEV
Compressor summary: The paper presents the first dynamic algorithm for facility location in high-dimensional spaces, achieving good quality and stable solutions with sub-linear update times.
https://openreview.net/forum?id=rtyqBfcg8j
Compressor summary: The paper proposes a new method to sample from unnormalized probability densities using a mean-field ODE and interacting particle systems that are gradient-free, closed-form, and require only sampling from a reference density and computing its ratio to the target density.
https://openreview.net/forum?id=rs8Sh2UASt
Compressor summary: Key points: - A flow-based generative modeling approach for learning and sampling protein conformational landscapes - AlphaFlow and ESMFlow are sequence-conditioned generative models of protein structure based on AlphaFold and ESMFold - Outperforms AlphaFold with MSA subsampling and captures conformational flexibility, positional distributions, and higher-order ensemble observables - Can diversify a static PDB structure faster than replicate MD trajectories Summary: The authors develop AlphaFlow and ESMFlow, generative models of protein structure that use flow methods to improve accuracy and diversity, and show their potential for simulating conformational flexibility and equilibrium properties.
https://openreview.net/forum?id=rqyXubsBhH
Compressor summary: This paper proposes a new design for decision support systems using prediction sets and online learning, which improves the performance by giving less agency to human experts.
https://openreview.net/forum?id=rmEgJ7bhuZ
Compressor summary: The paper introduces Gibbs Diffusion, a method that enables blind denoising by simultaneously sampling the signal and noise parameters using a conditional diffusion model and a Monte Carlo sampler.
https://openreview.net/forum?id=rkYOxLLv2x
Compressor summary: The paper evaluates large kernel convolutional neural networks (CNNs) and shows they can be as or more robust than vision transformers (ViTs), revealing novel insights into their source of robustness.
https://openreview.net/forum?id=rk4kmL8aOY
Compressor summary: The paper introduces RidCDR, a model for privacy-preserving cross-domain recommendation that uses embedding alignment to share knowledge without overlapping data.
https://openreview.net/forum?id=rfvgdfd1K9
Compressor summary: The text discusses the need for agency-preserving AI-human interactions that protect humans' long-term control, rather than just aligning AI systems with human intentions.
https://openreview.net/forum?id=reB9FFAaKw
Compressor summary: The paper studies how to collect data safely for policy evaluation in tabular MDPs and proposes a safe oracle algorithm called SaVeR.
https://openreview.net/forum?id=re6es2atbl
Compressor summary: The paper proposes a new assumption to analyze federated learning convergence rate under data heterogeneity and shows that it can reduce the impact of local Lipschitz constant and improve performance.
https://openreview.net/forum?id=rZD9hV0Bc4
Compressor summary: This paper proposes an efficient gradient-based algorithm for large-scale nonconvex Bi-Level Optimization problems in machine learning, with theoretical guarantees and experimental validation.
https://openreview.net/forum?id=rXnBvu5D7i
Compressor summary: The authors develop and test optimal dynamic programming methods to improve the scalability and performance of regression trees, a machine learning model that represents complex relationships.
https://openreview.net/forum?id=rVWsTjMW1m
Compressor summary: Coactive learning uses users' implicit edits to improve large language models without supervised training, enabling personalized LLMs.
https://openreview.net/forum?id=rU8o0QQCy0
Compressor summary: Machine learning methods can be valuable for causal inference and emulating simulations in natural sciences, but they also introduce unwanted biases in some cases.
https://openreview.net/forum?id=rTBR0eqE4G
Compressor summary: Component modeling helps to understand how machine learning models make predictions by breaking them down into their parts, and COAR is a tool that estimates the impact of each part on the prediction.
https://openreview.net/forum?id=rSfzchjIYu
Compressor summary: The paper presents UFGM, an unsupervised federated graph matching algorithm that uses graphlet features and trust region optimization to match node pairs across clients while preserving privacy.
https://openreview.net/forum?id=rReWhol66R
Compressor summary: The paper proposes a novel representation-based approach to measure the domain gap and filter data for cross-domain offline reinforcement learning, achieving superior performance with less target data.
https://openreview.net/forum?id=rPm5cKb1VB
Compressor summary: The authors propose a new GNN architecture and a fragmentation method with infinite vocabulary that improves expressiveness and performance on molecular property prediction compared to recent advances in higher-order GNNs, using the Fragment-WL test for theoretical analysis.
https://openreview.net/forum?id=rORsGuE2hV
Compressor summary: The paper proposes a new planning algorithm called highway value iteration network that combines value iteration with skip connections, exploration, and safety checks to enable effective long-term planning with deep neural networks.
https://openreview.net/forum?id=rMV86cAOh6
Compressor summary: The text introduces SCDM, a robust conditional diffusion model that generates realistic images from noisy semantic maps by stochastically perturbing the labels through Label Diffusion and using a class-wise noise schedule.
https://openreview.net/forum?id=rK6AZem0hX
Compressor summary: The paper proposes a method to learn consistent operations from latent embeddings by mapping them to a mirrored algebra on Euclidean space.
https://openreview.net/forum?id=rJxFvAs7pq
Compressor summary: The paper provides a comprehensive theoretical analysis of Actor-Critic algorithms, considering practical aspects such as multi-layer neural networks, Markovian sampling, continuous state-action spaces, last iterate performance, and global optimality, and shows their global convergence with sample complexity bounds.
https://openreview.net/forum?id=rJti61Uere
Compressor summary: A neural surrogate compiler uses a hypernetwork to generate efficient neural networks that mimic the behavior of programs, improving data efficiency and training speed compared to traditional methods.
https://openreview.net/forum?id=rJkGOARXns
Compressor summary: The paper investigates how transformer-based neural sequence models can learn different types of functions in-context by approximating them with neural networks and studying their pre-training and activation functions.
https://openreview.net/forum?id=rJVjQSQ8ye
Compressor summary: The paper proposes a method to train language models to generate long-form text with calibrated confidence statements, which helps users make better decisions based on the model's predictions.
https://openreview.net/forum?id=rIrpzmqRBk
Compressor summary: The paper proposes DRND, a modified RND algorithm that improves exploration in deep reinforcement learning by distilling a distribution of random networks and allocating bonuses more precisely.
https://openreview.net/forum?id=rIc9adYbH2
Compressor summary: Key points: - The paper proposes SEED-GNN, a GNN model editing method that is practical and effective. - The main challenge of GNN editing is sequential editing robustness, which is lacking in existing methods due to overfitting. - The paper also defines the task paradigm of GNN editing and hopes to inspire more research in this area. Summary: The paper introduces SEED-GNN, a novel method for editing graph neural networks (GNNs) that can handle multiple errors sequentially without overfitting, and formalizes the GNN editing task paradigm.
https://openreview.net/forum?id=rI6lxIX0uX
Compressor summary: The paper proposes a stochastic frame prediction model for learning image representations that captures uncertainty and temporal information, and shows its effectiveness on various video-based tasks.
https://openreview.net/forum?id=rHylzxK3HU
Compressor summary: The paper analyzes how class imbalance affects generalization curves for linear classifiers and proposes mixed under/oversampling strategies to improve performance.
https://openreview.net/forum?id=rGCvMARXkG
Compressor summary: PASOA is a new Bayesian experimental design method that uses contrastive estimation, stochastic optimization, and tempered SMC to balance information gain and accuracy in sequential design optimization and parameter inference.
https://openreview.net/forum?id=rEZ24oJhbn
Compressor summary: UPOCR is a simple and effective generalist model for pixel-level OCR tasks that unifies paradigms, architectures, and training strategies using vision Transformers and learnable task prompts.
https://openreview.net/forum?id=rADFNrIss3
Compressor summary: InstructZero is a method that optimizes soft prompts to create instructions for black-box LLMs, improving their performance on various tasks.
https://openreview.net/forum?id=r9rzU9QzPe
Compressor summary: The BiSHop model combines associative memory and attention mechanisms to handle non-rotational invariance and feature sparsity in deep tabular learning, achieving superior performance on real-world datasets with fewer hyperparameter searches.
https://openreview.net/forum?id=r9XICONppE
Compressor summary: The paper introduces a new relaxed solution to weighted low rank approximation (WLRA) that uses the weight matrix itself to reweight a low rank solution, achieving simple and efficient algorithms with provable approximation guarantees.
https://openreview.net/forum?id=r8k5JrGip6
Compressor summary: Superposition prompting is a novel method to improve efficiency and quality of large language models in retrieval-augmented generation by processing input documents in parallel paths and discarding irrelevant ones.
https://openreview.net/forum?id=r0qcGcFL4U
Compressor summary: The paper proposes a post-hoc method for routing tokens to specialized language models to improve zero-shot generalization, called PHATGOOSE.
https://openreview.net/forum?id=qz1Vx1v9iK
Compressor summary: The paper proposes a test-time adaptation method for resource-limited devices that uses input prompts, covariance matrix adaptation evolution strategy, and activation shifting to adapt models without backpropagation or weight changes.
https://openreview.net/forum?id=qwQVV5R8Y7
Compressor summary: The paper proposes a method to use pre-trained language models for time series forecasting by aligning their semantic space with temporal dynamics using token embeddings and learned prompts.
https://openreview.net/forum?id=qwKSTLbati
Compressor summary: The paper proposes a deep reinforcement learning model called Reinforced Leaf Sequencer to improve radiotherapy planning by optimizing leaf sequencing in a multi-agent framework.
https://openreview.net/forum?id=qstt2OguvM
Compressor summary: LaLiGAN is a novel generative model that can discover nonlinear symmetries in data and latent space, enabling applications like equation discovery and long-term forecasting.
https://openreview.net/forum?id=qqPL0DkcrI
Compressor summary: Sinusoidal positional encoding (SPE) is a new method that learns adaptive frequency features without hyperparameter tuning, improving fidelity and speed in various tasks such as 3D view synthesis, Text-to-Speech generation, and 1D regression.
https://openreview.net/forum?id=qoxuPshrZb
Compressor summary: The study examines how well vision-language models (VLMs) can estimate uncertainty across different settings and finds that temperature scaling helps improve calibration, even with a small amount of data.
https://openreview.net/forum?id=qoOt02l2WC
Compressor summary: The paper proposes a method to improve Integrated Gradients, a feature attribution technique for deep learning models, by aligning the path of attribution with the data manifold's geometry, resulting in more intuitive explanations and increased robustness to adversarial attacks.
https://openreview.net/forum?id=qmUbSAgz08
Compressor summary: The paper proposes a method to estimate uncertainty in machine learning predictions using multiple biased data sources, and shows its usefulness in predicting hospital length of stay for pediatric heart surgery patients.
https://openreview.net/forum?id=qklMNNub0H
Compressor summary: The text introduces a unified framework explaining how feedback alignment works in neural networks, improves its performance on multi-class tasks, and offers theoretical and empirical insights for better understanding and developing the method.
https://openreview.net/forum?id=qkhbyDqlNI
Compressor summary: This paper investigates how large language models can solve physical world problems through hierarchical reinforcement learning and exploration strategies.
https://openreview.net/forum?id=qjqlhWDcId
Compressor summary: The paper shows that transformers can learn sparse token selection and have better generalization than fully-connected networks.
https://openreview.net/forum?id=qg6AlnpEQH
Compressor summary: Cross-Task Linearity is a phenomenon where linearly interpolating weights of finetuned models starting from the same pretrained checkpoint results in similar features across tasks, suggesting neural networks act as approximate linear maps.
https://openreview.net/forum?id=qeFgvVVAJ2
Compressor summary: The memory-consolidated vision transformer (MC-ViT) extends the context for video understanding by fine-tuning pre-trained video transformers to attend to non-parametric memories, achieving state-of-the-art results on long-context tasks with fewer parameters.
https://openreview.net/forum?id=qbIKUfastZ
Compressor summary: Key points: - The paper proposes a reinforcement learning algorithm for episodic RMAB with unknown transitions, bandit feedback, and adversarial rewards - The algorithm has two main components: a biased adversarial reward estimator and a low-complexity index policy - The algorithm achieves $ ilde{\mathcal{O}}(H\sqrt{T})$ regret bound, which is the first to ensure $ ilde{\mathcal{O}}(\sqrt{T})$ regret for this problem setting Summary: The paper presents a novel reinforcement learning algorithm for sequential decision making problems with adversarial rewards and unknown transitions, using a biased reward estimator and a low-complexity policy, and showing an improved regret bound.
https://openreview.net/forum?id=qawwyKqOkj
Compressor summary: The paper proposes a bagging method for event-level loss functions in linear regression and GLMs, which reduces to one-dimensional clustering and improves model quality with adaptive homogeneous sampling and label differential privacy.
https://openreview.net/forum?id=qY63FnLuJ1
Compressor summary: The paper proposes a new pre-training strategy for protein models that combines residue and atom information, improving performance on downstream tasks like binding site prediction and function prediction.
https://openreview.net/forum?id=qY622O6Ehg
Compressor summary: The text discusses how strategically pausing decision updates in online reinforcement learning can improve performance by managing uncertainty, and provides theoretical and experimental evidence for this claim.
https://openreview.net/forum?id=qXoqV40imX
Compressor summary: The paper defines upper and lower bounds for neural network widths based on dataset complexity, explores how geometry affects network requirements, and develops an algorithm to infer dataset structure from trained networks.
https://openreview.net/forum?id=qTX1vxzs8b
Compressor summary: The paper studies how vision-language models can memorize and leak personal information from images during training, and proposes a countermeasure to prevent this issue.
https://openreview.net/forum?id=qRtM5EqE9l
Compressor summary: BLO-SAM is a model that improves semantic segmentation by optimizing prompt embeddings and using bi-level optimization to reduce overfitting and enable autonomous object segmentation.
https://openreview.net/forum?id=qQjUgItPq4
Compressor summary: DiCo is a method for controlling behavioral diversity in multi-agent systems by adjusting policy components, without changing the learning objective, and improving performance and sample efficiency.
https://openreview.net/forum?id=qOl2WWOqFg
Compressor summary: BiLLM is a 1-bit post-training quantization scheme that compresses pretrained LLMs while preserving their performance and efficiency.
https://openreview.net/forum?id=qOMQ0UGLYl
Compressor summary: DynSyn is a deep reinforcement learning algorithm that uses synergistic representations of muscle actuators derived from dynamical structures to improve motor control in high-dimensional, overactuated systems.
https://openreview.net/forum?id=qMG3OK7Xcg
Compressor summary: The paper proposes a novel method called Cluster-Aware Similarity for instance retrieval, which reduces misinformation propagation by diffusing similarity within local clusters instead of using pairwise instances.
https://openreview.net/forum?id=qLZ32oS7j2
Compressor summary: The paper proposes a decentralized algorithm to optimize user choice and service provider models in digital markets, and proves its convergence and effectiveness.
https://openreview.net/forum?id=qKL25sGjxL
Compressor summary: The paper introduces GiLOT, a method to measure and explain the impact of each word in large language models using Optimal Transport and token similarity.
https://openreview.net/forum?id=qIiPM5CbRY
Compressor summary: The paper studies a family of online decision problems that interpolate between learning with expert advice and multi-armed bandit, and provides tight minimax regret bounds and an optimal PAC algorithm for a special case called $\mathbf m$-BAI.
https://openreview.net/forum?id=qIOSNyPPwB
Compressor summary: The text discusses adversarial attacks on graph neural network explainers, showing that they can be easily manipulated by slightly perturbing the graph structure.
https://openreview.net/forum?id=qHt8FzPvU9
Compressor summary: This paper analyzes how weight precision affects ReLU neural networks in terms of number of neurons and preprocessing time, presenting an exponential algorithm to reduce neurons at the cost of precision, and showing that high precision alone doesn't help in reducing neurons.
https://openreview.net/forum?id=qGEEso256L
Compressor summary: The text describes a new method for predicting molecular properties using 2D and 3D representations of molecules integrated with a novel aggregation mechanism that is invariant under Euclidean transformations.
https://openreview.net/forum?id=qFILbkTQWw
Compressor summary: AnyTool is a powerful language model agent that uses over 16,000 APIs to address user queries and improves upon previous evaluation protocols with its own benchmark, AnyToolBench.
https://openreview.net/forum?id=qESG5HaaoJ
Compressor summary: The paper introduces a new neural network approach to compute eigenvalue decomposition efficiently and accurately using low-rank approximation and nesting techniques.
https://openreview.net/forum?id=qE4nkfyMYl
Compressor summary: The paper proposes a novel method to estimate discrete distributions in multivariate hypergeometric sampling when both population size and category sizes are unknown, using variational autoencoder framework and showing its applications in NLP and biology.
https://openreview.net/forum?id=qDw4FxMubj
Compressor summary: This paper proposes a sample-efficient model-based algorithm for learning robust equilibria in distributionally robust Markov games, where agents learn policies that perform well under various environmental uncertainties.
https://openreview.net/forum?id=qDUaH9xHVV
Compressor summary: Model-based MBR decoding improves text generation by using the model probability instead of a Monte Carlo estimate for hypothesis selection.
https://openreview.net/forum?id=qDAAMmGsGw
Compressor summary: The paper proposes ACM-MILP, a framework that generates better Mixed-Integer Linear Programming (MILP) instances by adaptively modifying constraints and modeling their interrelations.
https://openreview.net/forum?id=qAml3FpfhG
Compressor summary: The paper proposes strategies to reduce the number of evaluations needed for testing large language models on various benchmarks by using curated examples and releasing evaluation tools and smaller versions of popular benchmarks.
https://openreview.net/forum?id=q6fXuPLpao
Compressor summary: MLIP improves CLIP by using multiple domains and levels of supervision, token merging, and frequency transforms to enhance multimodal learning efficiency and performance.
https://openreview.net/forum?id=q5q59s2WJy
Compressor summary: The paper presents a Byzantine resilient algorithm for learning low-dimensional linear representation in a federated setting, with improved efficiency and security compared to existing methods.
https://openreview.net/forum?id=q5Bg858Hef
Compressor summary: The paper discusses concealed copyright infringement by generative models that use disguised copies of protected data for training and proposes methods to detect and prevent it.
https://openreview.net/forum?id=q3Bz1TVTq4
Compressor summary: The text discusses how users strategically choose whether to participate in predictive systems based on favorable outcomes, and proposes a framework for learning from such self-selected populations.
https://openreview.net/forum?id=q14AbM4kdv
Compressor summary: The paper proposes an algorithm to calibrate gradients for continual learning, reducing catastrophic forgetting and improving performance when historical data is limited.
https://openreview.net/forum?id=q0vILV7zAw
Compressor summary: The paper presents SDME, a multi-task network that combines sparse and dense feature matching for robust multimodal image registration, achieving excellent results on various datasets.
https://openreview.net/forum?id=q0lxAs5GGO
Compressor summary: TopDis is a novel method for learning disentangled data representations using a multi-scale topological loss that improves various disentanglement scores and works unsupervised, even for correlated factors of variation.
https://openreview.net/forum?id=pz4B2kHVKo
Compressor summary: The paper proposes a probabilistic routing method called PEOs that improves the efficiency of approximate nearest neighbor search in high-dimensional spaces using graph-based approaches.
https://openreview.net/forum?id=pwfcwEqdUz
Compressor summary: Key points: - The goal is to use Large Language Models (LLMs) to provide explanations for high-stakes event sequences. - The method builds on the temporal point process model and uses the likelihood function as a score for logic trees. - The approach combines an amortized EM learning framework with a GFlowNet generator for structured discrete variables. - The online setting extracts relevant rules from LLMs for each sequence in few iterations. Summary: The paper proposes a method to use LLMs for explanations of high-stakes event sequences, using a score based on the likelihood function and an EM learning framework with a GFlowNet generator.
https://openreview.net/forum?id=pvg1OdUtDQ
Compressor summary: DiNADO is an improved version of NADO for controllable language generation that addresses challenges like gradient vanishing, limited capacity, and allows combination with finetuning methods.
https://openreview.net/forum?id=puSMYmHmJW
Compressor summary: The paper proposes a mathematical model that represents the world as a hypergraph and studies how pre-training of foundation models can recover this structure using graph theory concepts.
https://openreview.net/forum?id=psup68MBvt
Compressor summary: DisCo-Diff is a new diffusion model that simplifies the learning process by adding discrete latents, improving performance in various tasks.
https://openreview.net/forum?id=pspyQm4ko0
Compressor summary: The BPA feature transform uses optimal transport optimization to efficiently and effectively represent high order relations between input features for various tasks like few-shot classification, clustering, and person re-identification.
https://openreview.net/forum?id=pp3v2ch5Sd
Compressor summary: Headless-AD is a single-trained model that can adapt to different action spaces without retraining or data collection.
https://openreview.net/forum?id=poEPRuNvM3
Compressor summary: Key points: - The paper proposes a framework for fair off-policy learning from observational data under different notions of fairness - The framework applies to actions or policy values as measures of fairness - The framework has theoretical guarantees and is tested on simulated and real-world data Summary: The paper presents a novel approach to learn fair decision rules from observational data in off-policy learning settings, with different fairness criteria and theoretical guarantees.
https://openreview.net/forum?id=po4NsL9KvX
Compressor summary: The paper presents a neural network architecture for learning tangent vector fields on surface manifolds that preserves intrinsic properties using a trainable vector heat diffusion module and vector-valued neurons.
https://openreview.net/forum?id=pmsPKIBAu6
Compressor summary: The text introduces a new framework for studying meta-learning methods using PAC-Bayesian theory, which allows more flexibility and directness in transferring knowledge between tasks, and demonstrates its effectiveness in both theory and practice.
https://openreview.net/forum?id=pmncWWkGMz
Compressor summary: The paper proposes agent-specific effects (ASE) to measure how an agent's action affects the outcome by influencing other agents and introduces counterfactual counterpart (cf-ASE) for identifying and estimating it.
https://openreview.net/forum?id=pmcusTywXO
Compressor summary: TopoOOD is a method for detecting out-of-distribution node instances on graphs that considers graph topology and neighborhood context, and shows improved performance over existing approaches.
https://openreview.net/forum?id=plXXbXjvQ9
Compressor summary: Deep ensembles become equivariant for all inputs and architectures with data augmentation, and this emergent property holds off-manifold in the infinite width limit.
https://openreview.net/forum?id=pktvuR7b5v
Compressor summary: EDDPM is an efficient denoising diffusion method that uses probabilistic masking to identify and skip redundant steps during training, improving inference efficiency without losing important information.
https://openreview.net/forum?id=pkUl39b0in
Compressor summary: The paper proposes a method to learn safe policies from expert demonstrations by inferring robust constraints that account for environmental differences, and evaluates it in continuous and discrete domains.
https://openreview.net/forum?id=piujJIF3zs
Compressor summary: Model Tailor is a method to reduce catastrophic forgetting in multi-modal language models by replacing a small portion of fine-tuned parameters and improving performance on both original and new tasks.
https://openreview.net/forum?id=piecKJ2DlB
Compressor summary: The paper explores using large multimodal models like GPT-4V as generalist web agents that can follow natural language instructions on websites, and proposes SEEACT to evaluate them on the MIND2WEB benchmark.
https://openreview.net/forum?id=phGHQOKmaU
Compressor summary: DiffStitch is a data augmentation method that connects low-reward trajectories with high-reward ones, improving offline reinforcement learning performance across different methods.
https://openreview.net/forum?id=pgI9inG2Ny
Compressor summary: This paper studies the optimal robust policy for deep reinforcement learning agents under state perturbations, proposes a new consistent adversarial robust deep Q-network algorithm, and shows its effectiveness in various benchmarks.
https://openreview.net/forum?id=pftXzp6Yn3
Compressor summary: The text introduces a new family of translation equivariant neural processes (TE-TNPs) that improve spatio-temporal modelling by leveraging symmetries in posterior predictive maps.
https://openreview.net/forum?id=pfnBLXgFVS
Compressor summary: Key points: - The paper proposes a general online algorithm for optimizing non-decomposable performance metrics based on confusion matrices - The algorithm is simple, efficient, and works for different types of classification problems - The algorithm achieves sublinear regret for concave and smooth metrics and performs well in experiments Summary: The paper presents an online algorithm that optimizes complex performance metrics using confusion matrices, with simplicity, efficiency, and low regret guarantees.
https://openreview.net/forum?id=pY2UpspnBB
Compressor summary: This paper proposes Distance-Aware Calibration (DAC), a simple and effective method to improve confidence calibration in vision-language models for open-vocabulary tasks using prompt learning.
https://openreview.net/forum?id=pXaEYzrFae
Compressor summary: The paper introduces DOMINO, a new algorithm that enforces constraints on large language models without sacrificing accuracy or speed.
https://openreview.net/forum?id=pVyOchWUBa
Compressor summary: This paper argues that non-identifiability in AR probabilistic models is a separate theoretical explanation for the desirable qualities of large language models, and discusses its relevance through three case studies.
https://openreview.net/forum?id=pTFud6SetK
Compressor summary: SelMatch is a novel dataset distillation method that effectively scales with the number of images per class and improves performance on image classification tasks.
https://openreview.net/forum?id=pSnhA7Em1P
Compressor summary: The paper proposes a subgoal-based learning framework that improves the performance of large language models in formal theorem proving by using demonstrative examples and diffusion models for organization.
https://openreview.net/forum?id=pQyoBWA146
Compressor summary: Split-Ensemble is a novel method that improves uncertainty estimation for deep learning models without extra OOD data or inference costs, using subtask-splitting and feature sharing.
https://openreview.net/forum?id=pPnkpvBeZN
Compressor summary: This paper proposes a topological augmentation framework called BAT that can effectively mitigate class imbalance bias in graph learning without reweighting or resampling.
https://openreview.net/forum?id=pPNMhdYMaz
Compressor summary: Key points: - The paper studies multi-agent reinforcement learning with adaptivity constraints - It proposes a policy elimination algorithm that achieves low regret and batch complexity - It proves lower bounds for such algorithms and extends to related problems Summary: The paper presents a near optimal algorithm and analysis for multi-agent reinforcement learning with adaptivity constraints, and shows its applicability to bandit games and reward-free MARL.
https://openreview.net/forum?id=pOgMluzEIH
Compressor summary: SILVER is a new optimization method for non-convex problems that reduces computation time and achieves optimal gradient complexity without needing multiple full gradients.
https://openreview.net/forum?id=pOJbk4Nzmi
Compressor summary: The paper proposes a new algorithm called ALEG for solving empirical Group Distributionally Robust Optimization (GDRO), which minimizes the maximal risk across multiple groups and outperforms existing methods.
https://openreview.net/forum?id=pLtuwhoQh7
Compressor summary: Mechanistic Neural Networks use a new block to learn differential equations and improve interpretability and efficiency in scientific data modeling.
https://openreview.net/forum?id=pFWmHUdJE5
Compressor summary: Key points: - Paper proposes O$(n)$-equivariant neurons with spherical decision surfaces for learning deep features under nd reflections and rotations - Neurons generalize to any dimension n and are called Deep Equivariant Hyperspheres - Network combines them using an invariant operator based on the relation between two points and a sphere - Approach outperforms competing methods on O$(n)$-equivariant benchmark datasets Summary: The paper introduces Deep Equivariant Hyperspheres, neurons that learn deep features under nd reflections and rotations with spherical decision surfaces. The network uses an invariant operator and shows superior performance on O$(n)$-equivariant tasks.
https://openreview.net/forum?id=pEWAcejiU2
Compressor summary: The paper proposes training language models to predict multiple future tokens simultaneously, improving sample efficiency, downstream capabilities, and inference speed for both code and natural language models.
https://openreview.net/forum?id=pDoAjdrMf0
Compressor summary: The paper investigates transfer reinforcement learning using successor features and generalized policy improvement, providing convergence analysis and generalization guarantees for this approach.
https://openreview.net/forum?id=pD9BTIDUoX
Compressor summary: The authors use (multi)modal DNNs to identify brain regions where multimodal integration occurs by predicting SEEG recordings from vision and language models.
https://openreview.net/forum?id=pBTLGM9uWx
Compressor summary: RAUCA is a method for generating effective adversarial camouflage against vehicle detectors, using a novel neural rendering component (NRP) and a multi-weather dataset.
https://openreview.net/forum?id=pAzDdYzEva
Compressor summary: QORA is an algorithm that uses a domain-agnostic object-based state representation to construct expressive models for various reinforcement-learning domains, achieving high generalization and interpretability with fewer observations than neural networks.
https://openreview.net/forum?id=pAyX8q1IIn
Compressor summary: The paper proposes new adaptive regularization strategies for stochastic weakly convex optimization that preserve the $\mathcal{O} ( 1 / \sqrt{K})$ convergence rate with a wide class of algorithms, using weak assumptions and showing efficiency and robustness in experiments.
https://openreview.net/forum?id=pAdI75JG3G
Compressor summary: The paper presents a novel framework for combinatorial multi-armed bandits with multivariant and probabilistically triggering arms, enhancing modeling power and achieving improved results in various applications like episodic reinforcement learning and probabilistic maximum coverage.
https://openreview.net/forum?id=pAPykbqUHf
Compressor summary: This paper develops a statistical theory for consistency models that speed up sample generation by merging steps in the diffusion process, achieving similar performance to diffusion models.
https://openreview.net/forum?id=pA2Q5Wfspp
Compressor summary: RL-CFR is a new RL method for dynamic action abstraction in IIEFGs that uses CFR for strategy derivation and achieves higher expected payoff than existing methods.
https://openreview.net/forum?id=p9SMltcfsu
Compressor summary: The paper proposes a method to reduce biases in black-box algorithms like neural networks by correcting non-linearities and handling scalar and tensor-valued predictions.
https://openreview.net/forum?id=p7gpooFIr3
Compressor summary: The paper proposes extensions of the two-metric projection framework to solve large scale nonconvex optimisation problems with nonnegativity constraints, achieving state-of-the-art convergence rates and competitive practical performance.
https://openreview.net/forum?id=p5FIjG9fbs
Compressor summary: This paper studies decision problems where agents receive some weak information about the hidden context and shows that current RL algorithms are not optimal for this setting, proposing a better algorithm with improved sample complexity.
https://openreview.net/forum?id=p225Od0aYt
Compressor summary: PRISE is a method that uses BPE to compress sequences and learn action abstractions for better skill learning in robotic manipulation tasks.
https://openreview.net/forum?id=p1kDNFs62o
Compressor summary: The paper presents a new Bayesian experimental design method for non-exchangeable data using Inside-Out SMC$^2$ algorithm and particle Markov chain Monte Carlo, which outperforms existing methods on dynamical systems.
https://openreview.net/forum?id=p0lKWzdikQ
Compressor summary: MEMORYLLM is a self-updatable language model that can memorize new text knowledge and maintain its performance after many updates.
https://openreview.net/forum?id=p0MGN0LSnx
Compressor summary: FedType is a framework for federated learning that uses proxy models to exchange information securely, efficiently, and without relying on public data.
https://openreview.net/forum?id=otuTw4Mghk
Compressor summary: This paper investigates why large language models encode high-level semantic concepts linearly, and shows that it is due to the loss function and gradient descent.
https://openreview.net/forum?id=or8BQ4ohGb
Compressor summary: InterpreTabNet uses a latent variable and GPT-4 to learn distinct and sparse feature masks for tabular data, improving interpretability and predictive performance.
https://openreview.net/forum?id=opkluZm9gX
Compressor summary: The paper studies a dynamic version of the attention matrix multiplication problem in large language models and provides an algorithm with efficient update and query time.
https://openreview.net/forum?id=opieUcKjPa
Compressor summary: The paper introduces a new family of distances called $k$-RPW that combines partial Wasserstein distance with robustness to outliers and faster convergence rate than existing measures, making it suitable for image retrieval tasks.
https://openreview.net/forum?id=oowQ8LPA12
Compressor summary: The paper presents a theoretical framework to analyze the performance of machine learning methods for database operations in dynamic datasets, showing when they can outperform non-learned alternatives.
https://openreview.net/forum?id=ooikIHLHCs
Compressor summary: The text discusses two complementary approaches to Explainable Artificial Intelligence (XAI) - human-oriented explanations (BLUE XAI) and model-oriented explanations (RED XAI), emphasizing the need for more methods in RED XAI to ensure AI safety.
https://openreview.net/forum?id=ooh8tkXKyR
Compressor summary: The paper introduces fault-tolerant PAC learning, a framework to identify robust machine learning models against random and adversarial faults, showing its sample complexity can vary depending on the type of fault.
https://openreview.net/forum?id=olbTrkWo1D
Compressor summary: The paper proposes a new continual learning algorithm, POCL, that optimizes past tasks' performance while maintaining the current task's performance, addressing the catastrophic forgetting challenge in machine learning.
https://openreview.net/forum?id=ojtddicekd
Compressor summary: Q-value regularized Transformer (QT) combines trajectory modeling with dynamic programming to improve offline reinforcement learning.
https://openreview.net/forum?id=oiY7yhyi6W
Compressor summary: The paper proposes new methods for off-policy evaluation without overlap or a well-specified model, using Lipschitz smoothness assumptions to provide sharp bounds and optimal estimators.
https://openreview.net/forum?id=ohH3sbUue2
Compressor summary: The text describes improved sampling methods for large data sets using sensitivity and subspace embeddings, achieving optimal complexity bounds with better guarantees than previous approaches.
https://openreview.net/forum?id=ohG9bVMs5j
Compressor summary: The paper proposes a new method to generate proxy graphs that help explain graph neural networks by preserving explanatory factors and adhering to the distribution of training data.
https://openreview.net/forum?id=ofXRBPtol3
Compressor summary: The paper proposes BSGAL, a new active learning algorithm that uses gradient cache to select batches of generated data for enhancing long-tailed instance segmentation tasks.
https://openreview.net/forum?id=odCl49tWA6
Compressor summary: The paper proposes a new algorithm (ME-$\mathcal{A}$) that improves robustness of neural networks in adversarial machine learning by ensuring uniform stability and mitigating the issue of robust overfitting.
https://openreview.net/forum?id=oaACFfNbXl
Compressor summary: The authors study when and why predictions are more valuable than expanding access or increasing intervention effectiveness in algorithmic decision making for public goods and welfare.
https://openreview.net/forum?id=oYltxxam2t
Compressor summary: VecKM is an efficient and accurate local point cloud geometry encoder that uses vectorized kernel mixtures and eliminates the need for grouping points into neighbors.
https://openreview.net/forum?id=oWYzIodyC4
Compressor summary: BWS is a universal data subset selection method that efficiently selects informative samples for training neural networks across various selection ratios using difficulty scores and kernel ridge regression.
https://openreview.net/forum?id=oUmXcewb83
Compressor summary: The text discusses a new method for finite-sample causal discovery that combines structured multiple testing and graph skeleton information, allowing efficient verification of graph structure.
https://openreview.net/forum?id=oTmQmaNkGn
Compressor summary: ERMI is a cognitive model that uses large language models to generate ecologically valid tasks and meta-learning to create rational agents, outperforming other models in explaining human behavior and achieving high accuracy on a benchmark task.
https://openreview.net/forum?id=oTYuORAMaP
Compressor summary: The paper proposes efficient stochastic approximation methods for minimizing excess risk in minimax optimization problems.
https://openreview.net/forum?id=oTD3WoQyFR
Compressor summary: The paper proposes a POMDP agent that uses meta-exploration techniques to gather relevant information for completing tasks in partially observed environments, outperforming prior methods when complex strategies are needed.
https://openreview.net/forum?id=oSOZ31ISBV
Compressor summary: The paper proposes a new test for conditional independence using an estimator based on the Von Mises entropy and kernel density estimation, which has better performance than existing methods in causal discovery for non-linear models and non-Gaussian continuous variables.
https://openreview.net/forum?id=oRLwyayrh1
Compressor summary: The paper proposes DRCT, a framework to improve the generalizability of image detectors for generated images by using high-quality diffusion reconstruction and contrastive learning, and introduces a large dataset with different diffusion models for evaluation.
https://openreview.net/forum?id=oOlooUu2Sb
Compressor summary: Key points: - The paper proposes a data selection method for offline imitation learning with imperfect demonstrations. - The method selects data based on resultant states, which uses dynamics information and extracts both expert and diverse behaviors. - The paper also presents a behavior cloning algorithm that leverages the selected data. - The method achieves state-of-the-art performance on complex and high-dimensional offline IL benchmarks. Summary: The paper introduces a resultant state-based data selection method and a behavior cloning algorithm for offline imitation learning with imperfect demonstrations, which outperforms existing methods on 20/21 benchmarks.
https://openreview.net/forum?id=oLfq1KKneW
Compressor summary: The paper proposes a framework using conditional residual energy-based models to improve molecule synthesis routes in drug discovery by considering criteria like costs, yields, and step count.
https://openreview.net/forum?id=oFDFGd9Age
Compressor summary: The Rashomon Effect means many good models exist for the same data, influencing various aspects of machine learning and its applications, especially for noisy tabular data problems.
https://openreview.net/forum?id=oDUJmNCV8D
Compressor summary: The GOUB model uses a generalized OU process to map from low-quality images to high-quality ones, achieving state-of-the-art performance in various image restoration tasks.
https://openreview.net/forum?id=oCI9gHocws
Compressor summary: Keyframe Identifier and Skill Annotator (KISA) is a method that uses visual-language representations to accurately and interpretably decompose unlabeled robotic manipulation demonstrations into keyframes and skills.
https://openreview.net/forum?id=oBYv73nOoA
Compressor summary: The paper introduces SPADE, a method that uses sample-targeted pruning to provide more accurate and interpretable image saliency maps and neuron visualizations for deep neural networks without affecting their behavior.
https://openreview.net/forum?id=oBP8vXFJNQ
Compressor summary: VideoPrism is a versatile video encoder that uses pretraining on large datasets to perform well on various video understanding tasks.
https://openreview.net/forum?id=o9uOuIwhZK
Compressor summary: The paper proposes using diffusion probabilistic schemes to improve learning of energy-based priors for multi-layer generative models, addressing the prior hole problem and increasing modelling expressivity.
https://openreview.net/forum?id=o8AaRKbP9K
Compressor summary: The paper explores how linear looped Transformers can learn and converge to fix-point iterative algorithms like preconditioned gradient descent in linear regression, using a novel theoretical analysis and experiments.
https://openreview.net/forum?id=o6N1Bqay0k
Compressor summary: The paper proposes a theoretical framework to quantify how deep learning models memorize spurious features unrelated to the task and reveals factors affecting this phenomenon.
https://openreview.net/forum?id=o5SVr80Rgg
Compressor summary: PairNet is a novel training strategy for individual treatment effect estimation that minimizes losses over pairs of examples based on their factual observed outcomes, achieving smaller generalization error than existing methods.
https://openreview.net/forum?id=o4HF3N6CZR
Compressor summary: The paper proves that ReLU neural networks with a width of $d+1$ can optimally approximate continuous functions over $[0,1]^d$ under $L^p$ norm for $p\in[1,\infty)$, and $d+11$ for the uniform norm.
https://openreview.net/forum?id=o2ND9v0CeK
Compressor summary: This paper shows how denoising diffusion models can be seen as a form of gradient descent, analyzes their convergence, and proposes a new sampler that achieves state-of-the-art performance in image generation.
https://openreview.net/forum?id=o1gS6MNAw8
Compressor summary: The authors propose a novel framework for concurrent visual and language planning, which outperforms separate approaches on various tasks, demonstrating the benefits of integrating vision and language.
https://openreview.net/forum?id=nxzXTLByXO
Compressor summary: BRAIn is a method that improves distribution matching techniques for language model alignment by reducing variance and generalizing the target distribution using Bayes' rule, leading to better performance in summarization and self-attention tasks.
https://openreview.net/forum?id=nvfZgdHtHc
Compressor summary: The study shows that using diverse data for training deep learning models improves their performance and robustness in accelerated MRI tasks without sacrificing in-distribution accuracy.
https://openreview.net/forum?id=nvHlHfjJPe
Compressor summary: The paper studies and generalizes tools for proving $H$-consistency bounds, a measure of how well regression algorithms perform, and derives new surrogate losses for adversarial regression with promising results.
https://openreview.net/forum?id=nsjfoziR5j
Compressor summary: The paper studies how to select representative sortition panels that reflect the population's opinion using metric distortion and compares two selection algorithms.
https://openreview.net/forum?id=nn5OPHom8t
Compressor summary: EVEREST is an efficient MVA approach for video representation learning that focuses on informative frames, reducing computation and memory requirements.
https://openreview.net/forum?id=nkOMLBIiI7
Compressor summary: SelfExtend is a method to extend the context window of language models without fine-tuning by using grouped and neighbor attention mechanisms.
https://openreview.net/forum?id=njwv9BsGHF
Compressor summary: LATS is a framework that combines language models and Monte Carlo Tree Search to create more capable autonomous agents for various decision-making tasks.
https://openreview.net/forum?id=njpTpkvUbO
Compressor summary: The paper introduces a new algorithm for learning in Constrained Markov Decision Processes, which achieves near-optimal regret bounds and performs better than existing methods in simulations.
https://openreview.net/forum?id=ngjmcfowtc
Compressor summary: The paper proposes a novel dynamical systems approach to fit latent variable models to data using optimal transport insights, which improves upon existing particle methods and other MLE algorithms.
https://openreview.net/forum?id=ngcZhfXCBW
Compressor summary: C3PO is a method that learns from high-level verbal feedback to adjust large language models without overgeneralizing the feedback to unrelated contexts.
https://openreview.net/forum?id=ndVXXmxSC5
Compressor summary: Key points: - Human motion taxonomies are useful for analysing how humans move and interact with their environment - They have a hierarchical structure but lack computational models that connect them to high-dimensional data - The proposed model uses hyperbolic embeddings and Gaussian processes to capture the taxonomy structure and learn from data - The model outperforms other methods and can generate realistic trajectories Summary: The paper proposes a novel model that uses hyperbolic embeddings and Gaussian processes to learn human motion taxonomies from data and generate realistic trajectories.
https://openreview.net/forum?id=nd47Za5jk5
Compressor summary: The paper proposes a graph-based method to unify relational and hierarchical inductive biases in deep learning for time series forecasting, using trainable graph pooling operators to learn the hierarchy from data and a differentiable reconciliation stage to balance constraints and predictions.
https://openreview.net/forum?id=ncjhi4qAPV
Compressor summary: The paper discusses the potential benefits and drawbacks of using non-private models pretrained on public datasets to improve differentially private machine learning.
https://openreview.net/forum?id=nbpwNmXTTw
Compressor summary: The paper introduces a new method for generating uncertainty bands that cover the entire path of random trajectories, useful for motion planning applications with unpredictable objects.
https://openreview.net/forum?id=nbOY1OmtRc
Compressor summary: The text discusses a random feature model that explains neural scaling laws in network training and generalization, predicting different power law exponents for performance with training time and model size, and showing convergence rates dependent on architecture and task.
https://openreview.net/forum?id=nYsh5GFIqX
Compressor summary: The paper presents a single end-to-end audio-visual large language model (av-LLM) called video-SALMONN, which improves speech understanding in videos and achieves significant accuracy gains on various audio-visual tasks.
https://openreview.net/forum?id=nYX7I6PsL7
Compressor summary: HAMLET is a graph transformer framework that uses modular input encoders to solve PDEs using neural networks, achieving robustness and adaptability across different domains and data scenarios.
https://openreview.net/forum?id=nUVForc3VP
Compressor summary: The paper proposes two improvements to existing two-timescale gradient methods for nonconvex-nonconcave minimax optimization, addressing instability and timescale separation challenges in overparameterized settings.
https://openreview.net/forum?id=nU1mtFDtMX
Compressor summary: The paper presents a methodology for assessing the economic rationality of LLMs as decision-making agents by surveying the literature, proposing a benchmark distribution, and conducting an empirical experiment.
https://openreview.net/forum?id=nTgzmXvuEA
Compressor summary: This paper demonstrates how predictive coding, a biological learning algorithm, can do causal inference by modifying its inference process without changing the causal graph and applying it to image classification tasks.
https://openreview.net/forum?id=nSGnx8lNJ6
Compressor summary: The paper proposes a method to improve offline reinforcement learning by combining discrete-time and continuous-time RL, using the value function's first derivative to better predict unvisited states.
https://openreview.net/forum?id=nP7Q1PnuLK
Compressor summary: THERMOMETER is a calibration approach that uses an auxiliary model to improve the accuracy and reliability of large language models for diverse tasks.
https://openreview.net/forum?id=nOyj26YdIQ
Compressor summary: The paper proposes minimal modifications to the transformer architecture to improve its performance on long-range tasks by incorporating smoothness and locality principles into the attention mechanism.
https://openreview.net/forum?id=nOjZfpLyh1
Compressor summary: BrainMixer is an unsupervised learning framework that leverages voxel-level activity and functional connectivity to represent the brain effectively, performing better than existing methods in cognitive tasks and neurological diagnosis.
https://openreview.net/forum?id=nMWxLnSBGW
Compressor summary: SHINE is a new method for protecting deep reinforcement learning agents from backdoor attacks by identifying and removing triggers in the agent's policy.
https://openreview.net/forum?id=nMN5hNZMQK
Compressor summary: The paper examines how different ways of parameterizing state-space models affect their memory learning abilities and introduces new reparameterization techniques to overcome the memory limitations of these models.
https://openreview.net/forum?id=nLgtHHBgl3
Compressor summary: The paper proposes MaskComp, a method that reconstructs incomplete objects using iterative generation and segmentation stages with a mask condition to improve image quality and refine the object mask.
https://openreview.net/forum?id=nLRKnO74RB
Compressor summary: The paper proposes a dynamic weight-ensembling module for merging task-specific Transformer models, which can adapt to different tasks and reduce parameter interference.
https://openreview.net/forum?id=nJzf3TVnOn
Compressor summary: The text proposes a new causal discovery algorithm that efficiently learns cause-and-effect relationships from interventional data using adaptive interventions and sampling history.
https://openreview.net/forum?id=nDps3Q8j2l
Compressor summary: The paper introduces FCNet, a new network that uses the frequency domain to improve data efficiency and inference speed in robotics reinforcement learning, outperforming Transformer on various tasks.
https://openreview.net/forum?id=nCZYRBK1J4
Compressor summary: The paper explores using reinforcement learning for online path planning in unknown environments and proposes a new map representation and reward function to improve coverage.
https://openreview.net/forum?id=nBPnmk6EeO
Compressor summary: Deep-Align is a novel framework that learns to solve the weight alignment problem of deep networks without requiring labeled data and improves both speed and quality of alignment compared to existing methods.
https://openreview.net/forum?id=nBGBzV4It3
Compressor summary: The paper proposes a new approach to optimize the sampling schedules of diffusion models for better quality outputs, using methods from stochastic calculus.
https://openreview.net/forum?id=nB6ERIud2y
Compressor summary: The paper proposes data integration methods for policy evaluation using experimental and historical data, with optimized weights and pessimistic principle to achieve robustness and efficiency in different reward shift scenarios.
https://openreview.net/forum?id=nAoiUlz4Bf
Compressor summary: Our approach provides robust certificates for message-passing neural networks using ReLU activation, mixed-integer optimization, and topology-based bounds tightening to ensure trustworthiness in the face of various graph attacks.
https://openreview.net/forum?id=nAbfF37H6t
Compressor summary: The authors propose a fast and exact method for simulating nonlinear resistive networks, enabling efficient training and reducing the simulation bottleneck.
https://openreview.net/forum?id=nACGn4US1R
Compressor summary: The paper suggests a way to prevent possible silent suffering in AI by limiting their access to memory or resetting it regularly, even without confirming their consciousness.
https://openreview.net/forum?id=n9pru4bJU9
Compressor summary: MNIST-1D is a minimalist, procedurally generated deep learning benchmark that enables various experiments and research on low-memory and low-compute platforms.
https://openreview.net/forum?id=n8g6WMxt09
Compressor summary: DeRa is a method for exploring different levels of regularization in aligned language models without retraining, allowing users to control alignment and improve efficiency.
https://openreview.net/forum?id=n3yYrtt9U7
Compressor summary: P$^2$INNs are a new extension of physics-informed neural networks that can efficiently model solutions of parameterized partial differential equations, improving accuracy and efficiency on benchmark problems.
https://openreview.net/forum?id=n3smZl8itR
Compressor summary: The paper proposes a method to simplify complex multi-player games by breaking them down into smaller subgames and applying Bellman's principle of optimality, improving scalability and efficiency.
https://openreview.net/forum?id=n2kq2EOHFE
Compressor summary: The paper investigates set membership estimation for unknown linear systems, providing the first convergence rate bounds and demonstrating its practical potential.
https://openreview.net/forum?id=n2eppIzHlL
Compressor summary: The paper proposes a new estimation strategy that combines regression and re-weighting methods to quantify the contribution of each causal mechanism in explaining the change in an outcome variable distribution.
https://openreview.net/forum?id=mzGtunvpJH
Compressor summary: The paper introduces D-iGPT, a new approach to learn visual representations from images by predicting semantic tokens and visible pixels with autoregressive pretraining using CLIP-based models.
https://openreview.net/forum?id=mz55Ox0Igz
Compressor summary: The paper proposes a new algorithm for offline linear bandits that minimizes Bayesian regret using efficient conic optimization solvers and shows its superiority over the maximum lower confidence bound approach.
https://openreview.net/forum?id=myCgfQZzbc
Compressor summary: The paper proposes a new method called Behavioral Eigenmaps (BeigeMaps) for learning representations in reinforcement learning agents from high-dimensional image observations that group similar states and improve policy performance.
https://openreview.net/forum?id=mxjB0LIgpT
Compressor summary: The paper discusses challenges and insights related to using evidential deep learning methods for quantifying uncertainty in ML systems.
https://openreview.net/forum?id=mv9beA1wDF
Compressor summary: The text discusses a problem in offline design optimization and proposes a new algorithm to improve surrogate models' accuracy by matching the latent gradient field in the data.
https://openreview.net/forum?id=muBJPCIqZT
Compressor summary: The authors show that using soft prompts can improve the performance of compressed large language models, making them more accessible without additional engineering efforts.
https://openreview.net/forum?id=mu7Er7f9NQ
Compressor summary: The paper introduces a new method to construct confidence sequences for multivariate stochastic processes using a general gambling framework, which improves upon existing methods in terms of tightness.
https://openreview.net/forum?id=mslTE1qgLa
Compressor summary: The paper introduces Major-Minor Mean Field Control (M3FC), a generalization of Mean Field Control that models many similar and few complex agents, and proposes an M3FMARL algorithm that approximates the policy gradient of the M3FC MDP and outperforms state-of-the-art methods in various scenarios.
https://openreview.net/forum?id=mrd4e8ZJjm
Compressor summary: The paper proposes a novel dynamics model for reinforcement learning that infers fine-grained causal structures using discrete latent variables, improving robustness and performance in downstream tasks.
https://openreview.net/forum?id=mphq2jMFLZ
Compressor summary: The paper investigates how two-layer neural networks can efficiently learn multiple kernel spaces and functions using mean-field Langevin dynamics in a two-timescale limit.
https://openreview.net/forum?id=moyG54Okrj
Compressor summary: The paper proposes a selective RAG framework that uses self-supervised learning to decide when to retrieve contexts for code completion, achieving better performance and efficiency than existing methods.
https://openreview.net/forum?id=mkbSXxovP5
Compressor summary: The text proposes two new decentralized optimization algorithms, snap-shot DSGT and accelerated snap-shot DSGT, which improve upon the existing DSGT method by using snapshot gradient tracking and achieving better convergence properties in different network topologies.
https://openreview.net/forum?id=mk8oRhox2l
Compressor summary: GliDe and CaPE are methods to speed up decoding of large language models by reusing cached keys and values and using confidence scores for token selection.
https://openreview.net/forum?id=mk3A5IUdn8
Compressor summary: Caduceus is a new family of DNA language models that can handle long-range interactions, bi-directionality, and reverse complementarity, improving performance on downstream tasks compared to larger models without these features.
https://openreview.net/forum?id=mjh7AOWozN
Compressor summary: NODESAFE improves GNNs' ability to detect out-of-distribution data in graphs by bounding negative energy scores and mitigating logit shifts, achieving better performance than previous methods.
https://openreview.net/forum?id=mhI5nc5QwX
Compressor summary: Key points: - The study observes low-rank structure in MHA sub-layer of Transformer, but not in FFN sub-layer - LoRAP is a novel method that combines low-rank approximation and structured pruning for LLMs - LoRAP outperforms previous methods on zero-shot perplexity and task classification Summary: The study proposes LoRAP, a new compression method for large language models that uses low-rank approximation and structured pruning, and shows its superior performance on various tasks.
https://openreview.net/forum?id=mggc3oYHy4
Compressor summary: The paper proposes a new attack on Decentralized Gradient Descent (D-GD) that allows an attacker to reconstruct private data of other users by exploiting the gossip averaging protocol.
https://openreview.net/forum?id=merZTLSdC9
Compressor summary: HIFIVE is a variational method for accurately estimating human preferences and causal effects from fragmented device identifiers in online behaviors.
https://openreview.net/forum?id=meItvvCO7X
Compressor summary: ToMo-UDA is a new method for detecting fetal structures in ultrasound images that uses topology and morphology knowledge to overcome challenges due to differences between institutions and overlapping structures.
https://openreview.net/forum?id=mcg6jppkwb
Compressor summary: The paper argues that tensor networks can enhance both sustainability and inclusivity in AI research by providing mathematical rigor and efficient compression.
https://openreview.net/forum?id=mbx2pLK5Eq
Compressor summary: A2Q+ is an improved quantization method for neural networks that avoids numerical overflow, reduces quantization error, and uses weight normalization.
https://openreview.net/forum?id=mbBehLOAqR
Compressor summary: The paper proposes a method to value data for machine learning models without knowing the validation dataset, using distributionally robust generalization error and model deviation as measures, and shows its effectiveness on real-world datasets.
https://openreview.net/forum?id=maVIKlGqr7
Compressor summary: The paper introduces HumanTOMATO, a framework for generating realistic whole-body motions from text descriptions, addressing the limitations of previous methods by using a hierarchical VQ-VAE and a Hierarchical-GPT model.
https://openreview.net/forum?id=mY93trX2Qz
Compressor summary: The paper proposes LoRS, a method to create synthetic data from image-text pairs for visual-language dataset distillation, which improves existing algorithms and focuses on modality correspondence.
https://openreview.net/forum?id=mXUDDL4r1Q
Compressor summary: The text discusses a fundamental problem in sequential decision making, presenting lower bounds and an algorithm for Reinforcement Learning (RL) from reachability specifications using expected conditional distance (ECD).
https://openreview.net/forum?id=mXLcbRBA8v
Compressor summary: The paper proposes an algorithm for building oblique classification trees that optimizes a loss function based on minimizing false negatives subject to a maximum false positive rate, which improves accuracy and interpretability in imbalanced datasets with class costs.
https://openreview.net/forum?id=mWV8NeU79e
Compressor summary: The text introduces Spider, a unified model that can understand and distinguish various context-dependent concepts in different domains, outperforming existing specialized models and enabling continuous learning.
https://openreview.net/forum?id=mUVydzrkgz
Compressor summary: ReaLchords is an online generative model for improvising chord accompaniment to user melody that uses reinforcement learning and a novel reward model for harmonic and temporal coherency.
https://openreview.net/forum?id=mUT1biz09t
Compressor summary: The text proposes using synthetic instructions to protect user privacy in language model applications, and shows that this method achieves high utility in both supervised and reinforcement learning settings.
https://openreview.net/forum?id=mUSPhG4uDW
Compressor summary: The paper introduces WEBLINX, a benchmark for conversational web navigation, and compares different models' performance in this task.
https://openreview.net/forum?id=mU7FfQT6VE
Compressor summary: PruNeRF is a framework to prune distracting objects from NeRF training data using 3D spatial consistency, segmentation, and depth-based reprojection.
https://openreview.net/forum?id=mNzkumTSVL
Compressor summary: Key points: - Decentralized FL is a serverless network where clients train local models separately - This may reduce model generalizability due to data and model heterogeneity among clients - DeSA introduces synthetic global anchors based on raw data distribution to facilitate knowledge transfer and regularization Summary: DeSA is a decentralized FL technique that uses synthetic global anchors to enhance the generalizability of local models trained separately by clients.
https://openreview.net/forum?id=mKYBMf1hHG
Compressor summary: The study explores the inconsistency of Data Shapley's performance in data selection tasks and identifies a class of utility functions where it works optimally, proposing a heuristic for predicting its effectiveness.
https://openreview.net/forum?id=mK6FB9xQ7v
Compressor summary: The paper compares the performance of a new optimizer, AdamQLR, which combines elements of first-order (Adam) and second-order (K-FAC) methods for deep learning optimization.
https://openreview.net/forum?id=mJhXlsZzzE
Compressor summary: This study analyzes the sample complexity and convergence of a shallow Graph Transformer for semi-supervised node classification, showing how self-attention and positional encoding improve generalization.
https://openreview.net/forum?id=mJGiFr8jLa
Compressor summary: The paper investigates challenges in training Physics-Informed Neural Networks (PINNs), compares different optimizers, and proposes a new second-order optimizer to improve their performance.
https://openreview.net/forum?id=mHIEOZtDDF
Compressor summary: The authors study various optimization techniques for tiny language models and achieve significant improvement in performance compared to baseline models.
https://openreview.net/forum?id=mGsF8Q0fGZ
Compressor summary: The paper proposes two transfer learning algorithms for functional linear regression, using RKHS distance to measure similarity and analyze dynamics, with empirical results on synthetic and real data.
https://openreview.net/forum?id=mDw42ZanmE
Compressor summary: MAIA is a system that uses neural models to automate understanding of other neural models' behavior, especially for vision tasks, by providing tools such as synthesizing inputs, computing activating exemplars, and summarizing results.
https://openreview.net/forum?id=mCzyRdDak5
Compressor summary: The paper presents two zero-shot audio editing techniques using pre-trained diffusion models, ZETA for text-based edits and ZEUS for semantically meaningful unsupervised modifications in music signals.
https://openreview.net/forum?id=mBc8Pestd5
Compressor summary: The paper introduces Reinforcer, a new sequence model for offline RL that maximizes returns and improves trajectory stitching.
https://openreview.net/forum?id=m8t1yzfBsJ
Compressor summary: The smooth min-max (SMM) network module is a simple modification of the existing min-max architecture that ensures monotonicity, improves training stability, and maintains good generalization performance in data-driven models.
https://openreview.net/forum?id=m8lCi7rG4u
Compressor summary: This paper investigates the causes of overfitting in adversarial training, finding that it's related to shortcuts formed in early layers of neural networks, and proposes a method to mitigate it by perturbing weights across layers.
https://openreview.net/forum?id=m5nB7ucXHT
Compressor summary: The study proposes a theory of representation learning in deep neural networks that applies to various architectures and activation functions, showing that some aspects of learning dynamics are similar across them.
https://openreview.net/forum?id=m4dO5L6eCp
Compressor summary: The paper proposes a lightweight and efficient smooth Tchebycheff scalarization approach for gradient-based multi-objective optimization with good theoretical properties and lower computational complexity.
https://openreview.net/forum?id=lwWV4Zl3h1
Compressor summary: The paper proposes a new method for uncertain predictions in machine learning that doesn't require data assumptions or learning rate tuning.
https://openreview.net/forum?id=lwTshcWlmB
Compressor summary: DfPO is a fine-tuning method for language models using reinforcement learning that prevents text degeneration and improves downstream task scores by masking KL divergence and using truncated advantage functions.
https://openreview.net/forum?id=luqH1eL4PN
Compressor summary: BiPE is a new positional encoding method for language sequences that combines intra-segment and inter-segment encodings to improve semantic information capture and extrapolation capabilities.
https://openreview.net/forum?id=ltzTHGFF5i
Compressor summary: Jetfire is an INT8 training method for transformers that optimizes memory access and maintains accuracy while speeding up pretraining and reducing memory usage.
https://openreview.net/forum?id=ltb2XaIr9p
Compressor summary: The paper analyzes two important optimization algorithms, gives their convergence rates, shows they stay near optimizers, provides an implementable version, and explores properties of a related method.
https://openreview.net/forum?id=lsavZkUjFZ
Compressor summary: CauDiTS is a framework for unsupervised domain adaptation of multivariate time series that disentangles causal patterns from correlations to improve classification reliability across domains.
https://openreview.net/forum?id=lsQnneYa8p
Compressor summary: The paper proposes a multi-task vehicle routing solver with mixture-of-experts and hierarchical gating, which improves generalization and performance on various vehicle routing problems.
https://openreview.net/forum?id=lsHZNNoC7r
Compressor summary: DistiLLM is a new knowledge distillation framework for language models that improves efficiency and performance by using a novel divergence loss and an adaptive off-policy approach.
https://openreview.net/forum?id=lrPrkWXqzd
Compressor summary: This paper proposes techniques to make distilling GANs from diffusion models more efficient, enabling real-time high-quality image editing on mobile devices with low training and storage costs.
https://openreview.net/forum?id=lrFwPeDdEQ
Compressor summary: Key points: - Paper introduces federated learning for online combinatorial optimization with bandit feedback - Transforms any offline resilient single-agent algorithm into an online multi-agent algorithm with better performance - Algorithm is communication-efficient and applies to stochastic submodular maximization Summary: The paper presents a federated learning framework that improves online combinatorial optimization with bandit feedback by transforming offline resilient algorithms into online multi-agent ones, achieving better regret bounds and communication efficiency.
https://openreview.net/forum?id=lqeVCc9zYq
Compressor summary: The paper proposes a score matching regularity (SMaRt) technique to improve GANs' ability to generate data that matches the real data manifold, especially for diverse and complex data.
https://openreview.net/forum?id=lpHjmPvxW1
Compressor summary: The paper proposes a backdoor defense framework for diffusion models called TERD, which can reverse triggers and detect backdoor inputs, achieving high security and adaptability.
https://openreview.net/forum?id=lon750Kf7n
Compressor summary: Density-Softmax is a sampling-free framework that improves uncertainty estimation and robustness by combining a density function with the softmax layer, achieving competitive results with fewer parameters and faster test time.
https://openreview.net/forum?id=lmiurzioja
Compressor summary: The paper proposes MoNA, a meta-learning method that learns data transformations to reduce modality gaps and improve cross-modality transfer.
https://openreview.net/forum?id=lm04PyXoEl
Compressor summary: The paper introduces new ways to measure and analyze how much one agent can affect another in multi-agent reinforcement learning, including approximation algorithms and empirical validation.
https://openreview.net/forum?id=limyQ1Kk0k
Compressor summary: The paper proposes a new method for predicting neuronal responses using neural networks and shows its superiority over Poisson models in inferring spike trains from retinal ganglion cells' recordings.
https://openreview.net/forum?id=lgcFX4VFrM
Compressor summary: The paper introduces a new type of generative models that can handle large and complex data sets by using ideas from mean-field theory and score-matching methods.
https://openreview.net/forum?id=leJGQCron2
Compressor summary: The paper analyzes the complexity of gradient methods for optimizing a Polyak–Łojasiewicz function with mean-squared smooth components and distributed on a network, and proposes an efficient decentralized method.
https://openreview.net/forum?id=lcX5GbDIi8
Compressor summary: The paper proposes a new Multi-Task Learning method that simultaneously groups tasks and trains a model in one-shot, using a differentiable Categorical distribution to prune task heads.
https://openreview.net/forum?id=laIOUtstMs
Compressor summary: PSBL is a meta-RL method that uses Bayesian inference to sample actions from the posterior distribution of the optimal policy, enabling in-context learning and improving performance on tasks with different distributions.
https://openreview.net/forum?id=lWy2lCTyJa
Compressor summary: The paper studies constrained and nonconvex min-max problems, improves first-order methods' complexity guarantees by using conic nonexpansiveness and relaxing inexactness levels, and analyzes a stochastic iteration method.
https://openreview.net/forum?id=lVQ4FUZ6dp
Compressor summary: The paper investigates how transformers can be trained to learn new sequential decision making tasks from few examples by using trajectory sequences with specific properties, and shows that larger datasets and more diverse tasks lead to better in-context learning.
https://openreview.net/forum?id=lT3W4AkyM7
Compressor summary: The paper proposes PLOT, an online tracking algorithm that learns a time-varying model of the target and uses it in receding horizon control, with theoretical results for non-stationary targets and real-world quadrotor demonstration.
https://openreview.net/forum?id=lQzmDFlsHX
Compressor summary: The paper proposes CoBalT, a concept balancing technique that mitigates spurious correlations in unsupervised object-centric learning without requiring human labeling of subgroups.
https://openreview.net/forum?id=lQIN9ZyMLz
Compressor summary: The paper proposes a causal inference framework to improve multi-label image classification by leveraging label correlations while mitigating overfitting issues caused by co-occurrence relationships.
https://openreview.net/forum?id=lQ3SEBH1gF
Compressor summary: GaussianPro is a novel method that improves 3D neural rendering by using progressive propagation and patch matching to densify 3D Gaussians, outperforming the traditional 3D Gaussian Splatting technique on large-scale scenes.
https://openreview.net/forum?id=lQ2o7JteMO
Compressor summary: The text discusses how to infer long-term causal effects of continuous interventions using short-term observations and doubly-robust estimators in offline reinforcement learning.
https://openreview.net/forum?id=lIYtJtpJR0
Compressor summary: Key points: - The paper studies the robustness of spiking neural networks (SNNs) against adversarial attacks by analyzing membrane potential perturbation dynamics. - The paper proposes a training framework with modified SNN neurons to reduce the mean square of membrane potential perturbation and enhance the stability of nonlinear systems. - The paper verifies the effectiveness of the framework on image classification task using Gaussian noise training and adversarial training. Summary: The paper improves the robustness of spiking neural networks against adversarial attacks by modifying their neurons to reduce perturbation dynamics and enhancing input-output stability.
https://openreview.net/forum?id=lHJFfDFbm6
Compressor summary: HelmFluid is a method that predicts fluid dynamics using the Helmholtz theorem, decomposing it into curl-free and divergence-free parts, and integrating them in multiple spatial scales.
https://openreview.net/forum?id=lGvIV4Bgsz
Compressor summary: The paper proposes a Convolution Bottleneck structure in CNNs, where early layers transform input into a few frequencies and channels, and later layers map back to outputs, explaining the common practice of down-sampling and how it affects function learning.
https://openreview.net/forum?id=lGZUvfP2ZF
Compressor summary: The text proposes PuTT, a method for learning compact and high-quality representations of visual data using tensor train upsampling, which improves compression, denoising, and image completion tasks.
https://openreview.net/forum?id=l9ga3iQuHt
Compressor summary: FGTS.CDB is a Thompson sampling algorithm for linear contextual dueling bandits with a new Feel-Good exploration term that achieves near minimax-optimal regret.
https://openreview.net/forum?id=l8GrPpsZfy
Compressor summary: Stochastic NGVI has a non-asymptotic convergence rate of $\mathcal{O}(\frac{1}{T})$ for conjugate likelihoods, similar to stochastic gradient descent, and likely converges faster in practice; however, for non-conjugate likelihoods, it optimizes a non-convex objective with no known global convergence rate.
https://openreview.net/forum?id=l7vQQi0I2d
Compressor summary: The text investigates whether there exist single-pass learning rules with maximum capacity in a certain class of rules, but finds such rules impossible using a linear program.
https://openreview.net/forum?id=l7shXGuGBT
Compressor summary: The paper proposes MATRIX, a social scene simulator that helps large language models align with human values by emulating realistic scenarios and fine-tuning the LLMs with simulated data, achieving better alignment than existing methods.
https://openreview.net/forum?id=l6Hef6FVd0
Compressor summary: PIPER is a novel hierarchical reinforcement learning method that uses preference-based learning and hindsight relabeling to learn from sparse rewards and improve performance in challenging tasks.
https://openreview.net/forum?id=l5lgbVR6BP
Compressor summary: This paper proposes a novel multiple kernel clustering framework that learns from expectation kernel matrices and provides theoretical guarantees for its performance.
https://openreview.net/forum?id=l5XQzNkAOe
Compressor summary: The text introduces TravelPlanner, a new planning benchmark for testing language agents' abilities in complex real-world scenarios, showing that current language models struggle to handle it.
https://openreview.net/forum?id=l4ZjeDDnu9
Compressor summary: The paper proposes a novel algorithm for sampling rows in a matrix based on their norm when the data is presented as a turnstile data stream, which can improve subsampling constructions for regression problems with low overhead.
https://openreview.net/forum?id=l4H7Hv7LhJ
Compressor summary: The paper proposes a method to build decision trees that can handle hierarchical groups and generalize well with few samples.
https://openreview.net/forum?id=l1YbS3qkdk
Compressor summary: The paper proposes a new method to test causal relationships over continuous variables without parametric assumptions, using discretization and a novel test statistic.
https://openreview.net/forum?id=l0OGoZPZuC
Compressor summary: CDAM is a novel associative memory model that uses graph structures to link memory patterns and can handle auto- and hetero-association with anti-Hebbian learning rules for various applications.
https://openreview.net/forum?id=kzz0kn546b
Compressor summary: The text explains how neural network-based active learning works by prioritizing samples with yet-to-be-learned features, and shows that both uncertainty-based and diversity-based query criteria achieve this goal.
https://openreview.net/forum?id=ksph9pkEDc
Compressor summary: Selective mixup improves generalization by non-randomly selecting pairs, which resamples data to uniform class distribution, an effect explained by regression toward the mean.
https://openreview.net/forum?id=ks8qSwkkuZ
Compressor summary: The paper proposes a safe reinforcement learning method that considers goal achievement and reduces ineffective exploration by using a feasible reachable function and policy iteration.
https://openreview.net/forum?id=kpDd2HCBka
Compressor summary: The paper proposes methods to make online Monte Carlo estimators more efficient and unbiased by learning a tailored behavior policy from offline data.
https://openreview.net/forum?id=knhbhDLdry
Compressor summary: The paper explores how using in-distribution (ID) labels can improve out-of-distribution (OOD) detection in machine learning by analyzing data separability with a graph-theoretic approach.
https://openreview.net/forum?id=knZ4NYzGUd
Compressor summary: Bayesian Knowledge Distillation (BKD) is a method that connects KD to Bayesian modeling, providing insight into its working mechanism and tools for measuring student model uncertainty.
https://openreview.net/forum?id=kn2xp8UOvQ
Compressor summary: The paper introduces BiLipNet, an invertible neural network that can control its output sensitivity and input distinguishability, and PLNet, a scalar-output network based on BiLipNet and quadratic potential, which can learn non-convex losses with global minimum.
https://openreview.net/forum?id=kmugaw9Kfq
Compressor summary: The paper introduces EPNN, a novel message-passing framework for spectral invariant GNNs, and analyzes their expressiveness compared to other architectures and subgraph GNNs.
https://openreview.net/forum?id=klKk9ETAyU
Compressor summary: The paper proposes a new online-to-non-convex framework that uses heavy-tailed gradients and gradient clipping to find stationary points with high probability, improving on existing results for smooth and non-smooth objectives.
https://openreview.net/forum?id=kkqIEp2bRa
Compressor summary: The paper proposes two differentially private algorithms for adapting predictions from public data to private data with similar performance as non-private methods.
https://openreview.net/forum?id=kfpe7Dg23G
Compressor summary: The text proposes a new optimization-based approach to improve spiking neural networks by approximating subgradient methods and expanding their nonlinearity support, achieving state-of-the-art conversion from artificial neural networks.
https://openreview.net/forum?id=kc4dZYJlJG
Compressor summary: FedRC is a new algorithm that helps protect privacy in machine learning by handling multiple types of data shifts among clients using clustering and bi-level optimization.
https://openreview.net/forum?id=kao5hRX9YA
Compressor summary: BAT is a model that combines spatial sound perception and natural language reasoning to navigate and interpret in-the-wild spatial sounds using synthesized audio data and a novel spatial audio encoder.
https://openreview.net/forum?id=kZbTkpnafR
Compressor summary: The paper investigates how Transformers learn to predict the next token in a sequence by training on a simple first-order autoregressive task and shows that they do so through an in-context autoregressive learning procedure, where they learn orthogonal matrices to capture data patterns.
https://openreview.net/forum?id=kZKopcDp2q
Compressor summary: The text presents a novel way to study optimizers in neural networks using hyperbolic geometry and dynamical systems tools, focusing on their long-term behavior and stability.
https://openreview.net/forum?id=kZBCFQe1Ej
Compressor summary: The paper proves that Distributional Reinforcement Learning can achieve better instance-dependent bounds in both online and offline RL than previous methods, and demonstrates its effectiveness in contextual bandits.
https://openreview.net/forum?id=kZArjKc64o
Compressor summary: IBSF is a novel fingerprinting method for detecting tampering with DNN models using only top-1 labels, which maximizes the partial Shannon entropy of selected categories near decision boundaries.
https://openreview.net/forum?id=kXde6Qa6Uy
Compressor summary: The text proposes a new framework for estimating graphical model parameters from incomplete data using optimal transport, without making unrealistic assumptions or variational approximations, and shows its effectiveness and robustness in experiments.
https://openreview.net/forum?id=kXHgEYFyf3
Compressor summary: The paper introduces R2E, a framework that converts GitHub repositories into test environments for evaluating AI coding assistants using program analysis and large language models.
https://openreview.net/forum?id=kVgpa1rfLO
Compressor summary: The paper introduces a new unsupervised federated learning algorithm (FedGrEM) for mixture models, with a comprehensive finite-sample theory that compares its performance to local single-task learning and other federated EM algorithms.
https://openreview.net/forum?id=kUm9iuvwIQ
Compressor summary: Slicedit is a method for text-based video editing that uses a pre-trained image synthesis model to process spatial and spatiotemporal slices, achieving better results than existing methods.
https://openreview.net/forum?id=kUj9b2CezT
Compressor summary: The paper proposes a novel method, C$^2$GAM, to generate missing data from different environments and address collider bias in observational studies.
https://openreview.net/forum?id=kTaX87Zn6M
Compressor summary: The paper proposes techniques to accelerate and preserve accuracy in pre-training large transformers using 2:4 sparse matrix multiplication on NVIDIA Ampere GPUs.
https://openreview.net/forum?id=kRxCDDFNpp
Compressor summary: Best-fit Packing is a method that optimizes document packing for large language models, reducing truncations and improving coherence and performance.
https://openreview.net/forum?id=kRv0WPJd00
Compressor summary: The variational Schr"odinger diffusion model (VSDM) improves scalability and efficiency of transportation plans in diffusion models by linearizing forward score functions with variational inference and optimizing backward scores without simulation-based losses.
https://openreview.net/forum?id=kQwSbv0BR4
Compressor summary: DFMs are a new flow-based generative model that can handle discrete and continuous data, achieving state-of-the-art results in protein co-design tasks.
https://openreview.net/forum?id=kQ1dwuheR0
Compressor summary: PPAE is a text-guided, training-free method for precise audio editing using cross-attention maps and a hierarchical pipeline.
https://openreview.net/forum?id=kOczKjmYum
Compressor summary: MusicFlow is a text-to-music model that uses flow matching networks and masked prediction to generate high-quality music from text descriptions efficiently.
https://openreview.net/forum?id=kMBvZ40Iu9
Compressor summary: The paper proposes a method to reduce the computational cost of numerical simulations using differentiable physics, $k$-means clustering, and stochastic minimization for coarsening unstructured grids.
https://openreview.net/forum?id=kLiDMGJKx1
Compressor summary: The Outlier-Efficient Modern Hopfield Model improves associative memory retrieval and attention in large transformer-based models, reducing kurtosis and infinity norm of model outputs.
https://openreview.net/forum?id=kLZZWvqlEm
Compressor summary: Key points: - Autoencoders are useful for generative modeling and representation learning - Structural constraints like conditional independence can improve latent variable invariance - Wasserstein autoencodters (WAEs) can easily incorporate such constraints - StrWAEs are a principled way of penalizing autoencoders to impose structural constraints - StrWAEs show promising results on various tasks Summary: The paper introduces StrWAEs, a flexible and principled way of imposing structural constraints on autoencoders using Wasserstein autoencoders, which improve latent variable invariance and perform well on several tasks.
https://openreview.net/forum?id=kKWjZoaRLv
Compressor summary: The GPGVAE model learns the strategic interactions and network structures from observed actions in network games using a spectral GNN encoder, a data-dependent gated prior, and a Transformer mixture of Bernoulli encoder.
https://openreview.net/forum?id=kIh7GJmRfD
Compressor summary: ATraDiff is a generative diffusion model that adapts to different trajectory lengths and distribution shifts, improving online RL performance using synthetic trajectories from offline data.
https://openreview.net/forum?id=kIHIA6Lr0B
Compressor summary: The paper introduces a hydraulics-informed graph neural network for flood simulation that incorporates physics domain knowledge and performs well on complex topography and sparse precipitation data.
https://openreview.net/forum?id=kHjOmAUfVe
Compressor summary: RoboDreamer is an innovative method that learns a compositional world model by factorizing video generation, enabling generalization of language instructions for realistic plan synthesis and multimodal goals in robotic decision-making.
https://openreview.net/forum?id=kGXUL6qGso
Compressor summary: The paper presents a new method for sequentially selecting features in health care applications that minimizes costs and improves diagnostic performance using an oracle based approach.
https://openreview.net/forum?id=kAfYYg6PX8
Compressor summary: L-MAC is an interpretation method for audio classifiers that generates binary masks to highlight relevant parts of audio signals, which are more faithful and preferred by users than other methods.
https://openreview.net/forum?id=kAIkYOE5pV
Compressor summary: The paper proposes a novel Spatio-Temporal Circuit (STC) model for spiking neural networks to improve their ability to handle complex, dynamic spatio-temporal prediction tasks by incorporating autaptic synapses and two learnable adaptive pathways.
https://openreview.net/forum?id=kAFevjEYsz
Compressor summary: Key points: - Existing methods improve adversarial robustness but only on in-distribution (ID) data - OODRobustBench is a benchmark to assess out-of-distribution (OOD) robustness using 23 dataset shifts and 6 threat models - Adversarial robustness suffers from severe OOD generalization issue - ID robustness correlates strongly with OOD robustness in a positive linear way - Existing methods are unlikely to achieve high OOD robustness and novel methods are needed Summary: The paper introduces OODRobustBench, a benchmark to measure OOD adversarial robustness using various shifts and models, and shows that existing methods have poor OOD performance and require new solutions.
https://openreview.net/forum?id=k7G4N1x7f9
Compressor summary: The paper proposes a new version of Sharpness-Aware Minimization (SAM) called BiSAM, which improves the performance of network perturbations using a bilevel optimization approach and a novel lower-bound surrogate loss.
https://openreview.net/forum?id=k5ncz7TIPX
Compressor summary: DDDMs generate realistic images quickly using few-step sampling and Pseudo-LPIPS, outperforming GANs and distillation-based models on benchmark datasets.
https://openreview.net/forum?id=k2dVVIWWho
Compressor summary: This paper studies the privacy of decentralized learning with random walk algorithms using a new differential privacy variant, and shows that it can improve privacy compared to gossip algorithms for nearby nodes.
https://openreview.net/forum?id=k2axqNsVVO
Compressor summary: SDEA is a method for Byzantine-robust aggregation in heterogeneous federated learning that uses a random public dataset and learns aggregation weights to distinguish benign from malicious clients.
https://openreview.net/forum?id=k1JXxbpIY6
Compressor summary: The authors investigate how large language models perform on arithmetic word problems, comparing their cognitive biases to those of children, and find that LLMs mimic human biases in understanding text and planning solutions, but not in executing arithmetic expressions.
https://openreview.net/forum?id=k1J2GbamLi
Compressor summary: The paper proposes a method to find cohorts within negative samples in healthcare analytics using data Shapley values and manifold learning, which can reveal insights on diseases and related conditions.
https://openreview.net/forum?id=k10805cgak
Compressor summary: The paper proposes a new cutting plane method for solving integer linear programs by removing some existing constraints, which improves the performance compared to adding new ones.
https://openreview.net/forum?id=jzHmElqpPe
Compressor summary: Key points: - Decision making needs perception, memory, and reasoning for optimal policies - Conventional approaches have limitations in sample efficiency and generalization - Foundation models in language and vision can adapt to diverse tasks faster - The authors propose foundation agents as a new learning paradigm inspired by LLMs - They outline the roadmap of foundation agents from data collection/generation, pretraining, adaptation, and alignment with LLMs - They identify critical research questions and trends for foundation agents with real-world use cases Summary: The paper proposes foundation agents, a new learning paradigm for decision making inspired by large language models, that can adapt faster and better to diverse tasks with data collection/generation, pretraining, adaptation, and alignment.
https://openreview.net/forum?id=jxvqvZLBuU
Compressor summary: RNAFlow is an AI method for designing RNA structures and sequences by combining a denoising network with an RNA inverse folding model that considers conformational flexibility.
https://openreview.net/forum?id=jw2f9v59g0
Compressor summary: BOOT is an efficient data-free distillation technique for improving the speed and quality of image generation using diffusion models without requiring additional data or computationally expensive processes.
https://openreview.net/forum?id=jvVWPtJYbc
Compressor summary: The paper proposes a new method to estimate velocity fields for Wasserstein Gradient Flow, which improves the accuracy of particle movement and applies it to domain adaptation and missing data imputation tasks.
https://openreview.net/forum?id=jsmaWEdx9g
Compressor summary: Key points: - propose a new optimization model for clustering applications called sum-of-minimum optimization - develop efficient algorithms inspired by k-means and Lloyd's algorithm - prove a tight bound and convergence rate for the algorithms - show numerical results on multiple tasks Summary: The paper introduces sum-of-minimum optimization, a novel clustering model with efficient algorithms based on k-means and Lloyd's algorithm, and demonstrates its effectiveness on various tasks.
https://openreview.net/forum?id=jsKr6RVDDs
Compressor summary: The text discusses the challenges of defining and measuring diversity in machine learning datasets, and suggests using principles from measurement theory to improve dataset construction.
https://openreview.net/forum?id=jrHUbftLd6
Compressor summary: FedMBridge is a new method for multimodal federated learning that uses a hypernetwork to handle different client architectures and data types while protecting privacy.
https://openreview.net/forum?id=jrE7geZekq
Compressor summary: The paper proposes PGODE, a new method for modeling multi-agent dynamical systems using prototype decomposition and graph ODEs, which improves generalization under system changes and outperforms baselines in various scenarios.
https://openreview.net/forum?id=jr0W36wOBx
Compressor summary: The paper proposes a new conformal regression method to improve survival analysis models' calibration without compromising their ability to rank subjects accurately.
https://openreview.net/forum?id=jnps5YwNlU
Compressor summary: The paper proposes eP&R, a new evaluation metric for deep generative models that uses hubness-aware sampling to reduce computational costs while maintaining accuracy.
https://openreview.net/forum?id=jn2iTJas6h
Compressor summary: The paper presents a time-series foundation model for forecasting that performs well without any fine-tuning on various public datasets using a large pretrained attention model with input patching.
https://openreview.net/forum?id=jmmji1EU3g
Compressor summary: The In-context Decision Transformer (IDT) improves offline reinforcement learning efficiency by using a hierarchical structure inspired by human decision-making to reconstruct high-level decisions instead of low-level actions, achieving state-of-the-art results in long-horizon tasks and reducing evaluation time significantly.
https://openreview.net/forum?id=jklD0TV5Hw
Compressor summary: Seesaw is a novel neural architecture search method that leverages more linear computations and nonlinear result reuse for privacy-preserving machine learning, achieving better accuracy and lower latency than existing methods.
https://openreview.net/forum?id=jhWSzTO0Jl
Compressor summary: The paper proposes a method to create post-hoc part-prototype networks that can explain both where and what a model looks for in an image, while maintaining performance and providing more faithful explanations.
https://openreview.net/forum?id=jh7FDDwDBf
Compressor summary: The paper proposes plug-in performative optimization, a method that uses possibly misspecified models to reduce performative risk and improve prediction in situations where the feedback affects future observations.
https://openreview.net/forum?id=jdRIaUu3xY
Compressor summary: BBox-Adapter is a novel method for adapting black-box LLMs to specific tasks using ranking-based NCE loss and online data sampling, improving performance and reducing costs.
https://openreview.net/forum?id=jbPc3pW6sC
Compressor summary: The paper proposes online VSMC, a method for efficient and accurate parameter estimation and latent state inference in state-space models using stochastic gradient approximation.
https://openreview.net/forum?id=japBn31gXC
Compressor summary: The text discusses how offline reinforcement learning algorithms struggle with diverse data from different tasks, and suggests using larger policy sizes to improve performance.
https://openreview.net/forum?id=jaJxpKkBcL
Compressor summary: The paper studies how cardinality sketches perform in adaptive settings and reveals their vulnerabilities, showing an attack that exploits simple non-adaptive queries to generate adversarial inputs.
https://openreview.net/forum?id=jZVen2JguY
Compressor summary: The Flexible Vision Transformer (FiT) is a new transformer architecture that can generate images with unrestricted resolutions and aspect ratios by conceptualizing images as sequences of dynamically-sized tokens, thus promoting resolution generalization and eliminating biases.
https://openreview.net/forum?id=jZEY5SxbL4
Compressor summary: Key points: - CraFT is a method for fine-tuning black-box VLMs using input prompts and output predictions only - CraFT consists of two modules: prompt generation and prediction refinement, with an auxiliary loss to ensure consistency - CraFT outperforms white-box methods on few-shot classification with less queries, faster training, and less memory Summary: CraFT is a novel approach for fine-tuning black-box VLMs without accessing their parameters. It uses prompt generation and prediction refinement modules to improve few-shot classification performance, efficiency, and memory footprint.
https://openreview.net/forum?id=jXn1qIcjyG
Compressor summary: The paper introduces conditional finetuning, a method that allows language models to learn useful knowledge from a corpus while avoiding unnecessary biases, leading to improved performance on downstream tasks and lifelong learning.
https://openreview.net/forum?id=jWHU4b7Yk6
Compressor summary: SyCoCa is a method that improves multimodal alignment between language and vision by introducing bidirectional interactions on global and local representations using textual and visual cues.
https://openreview.net/forum?id=jVXJdGQ4eD
Compressor summary: MagicPose is a diffusion-based model that can generate realistic human images with controlled poses and facial expressions while preserving the identity, using a two-stage training strategy to disentangle appearance and motions.
https://openreview.net/forum?id=jU6iPouOZ6
Compressor summary: The paper introduces a novel framework called Vision to Audio and Beyond (VAB) that learns representations and generates modalities within latent spaces for various audio-visual tasks.
https://openreview.net/forum?id=jTn4AIOgpM
Compressor summary: DDVI is a novel method that uses denoising diffusion SDEs and neural networks to infer the posterior distribution of sparse inducing points in deep Gaussian processes, improving efficiency and reducing bias.
https://openreview.net/forum?id=jS3CMHtYJD
Compressor summary: The paper presents new coresets for classification problems that have smaller size, distributional input support, and various loss function applications.
https://openreview.net/forum?id=jRX6yCxFhx
Compressor summary: Open foundation models have benefits and risks related to innovation, power distribution, and transparency, but their misuse risk is unclear due to insufficient research and varying assumptions in past work.
https://openreview.net/forum?id=jQA5iutPzd
Compressor summary: The paper studies how well deterministic methods solve inverse problems in imaging and shows that they need higher Lipschitz constants for better perceptual quality and consistency, making them more vulnerable to attacks.
https://openreview.net/forum?id=jQ92egz5Ym
Compressor summary: The paper proposes IR-QLoRA, a method to improve the accuracy of quantized large language models for deployment on resource-constrained devices by retaining original information through statistics-based and finetuning-based technologies.
https://openreview.net/forum?id=jPaEOH56JB
Compressor summary: The paper proposes a geometric diffusion model with learnable divergence fields and causal inference to improve generalization for interdependent data, and provides three model instantiations based on GCN, GAT, and Transformers.
https://openreview.net/forum?id=jP8mf34iCW
Compressor summary: The paper proposes a new subset selection algorithm for active learning that optimizes batch acquisition directly on the combinatorial space using sequential greedy sampling and improves efficiency in solving expensive multi-objective combinatorial optimization problems.
https://openreview.net/forum?id=jP1zeEqHli
Compressor summary: The paper proposes Lookahead-Replicate, a new reinforcement learning algorithm that maintains function space equivalence between online and target networks, leading to better deep RL performance on Atari games.
https://openreview.net/forum?id=jOlO8t1xdx
Compressor summary: The paper proposes an efficient generative model that creates long-form, variable-length stereo audio from text prompts using latent diffusion and convolutional autoencoders, outperforming existing models in quality and structure.
https://openreview.net/forum?id=jNab9mXEyj
Compressor summary: The paper introduces a new type of neural network that combines partial stochasticity and infinite depth to improve efficiency, expressivity, and performance on various tasks.
https://openreview.net/forum?id=jNM4imlHZv
Compressor summary: The paper shows how gradient descent on a simplified two-layer transformer learns latent causal structure from in-context learning tasks, and proves that the attention matrix encodes mutual information between tokens.
https://openreview.net/forum?id=jM9A3Kz6Ki
Compressor summary: Compound returns, a weighted average of step returns, can reduce variance in reinforcement learning methods, improving their sample efficiency.
https://openreview.net/forum?id=jKnW7r7de1
Compressor summary: BetterV is a Verilog generation framework that uses fine-tuned large language models and generative discriminators to create correct and optimized Verilog code for various electronic design automation tasks.
https://openreview.net/forum?id=jKYyFbH8ap
Compressor summary: SAFIM is a new benchmark to evaluate Large Language Models on code Fill-in-the-Middle tasks, focusing on syntax-aware completions and providing a robust framework for accurate and fair comparisons.
https://openreview.net/forum?id=jKUWlgra9b
Compressor summary: ERQ is a two-step method that reduces quantization error in vision transformers by strategically updating weights with full-precision and refining rounding directions of quantized weights.
https://openreview.net/forum?id=jJmGl01S4l
Compressor summary: The paper explains how spikes in training loss during neural network training with stochastic gradient descent (SGD) are related to improved feature learning and generalization by increasing alignment with the true predictor's average gradient outer product.
https://openreview.net/forum?id=jJLcXGB2uA
Compressor summary: The paper studies how to schedule jobs efficiently when only some predictions are available, and proposes an algorithm that is robust, consistent, and smooth under this restriction.
https://openreview.net/forum?id=jJ9BoXAfFa
Compressor summary: This paper proposes using executable Python code as a unified action space for large language models, improving their performance on agent-oriented tasks and enabling them to perform sophisticated actions like model training.
https://openreview.net/forum?id=jEoIkNkqyc
Compressor summary: X-MDPT is a novel pose-guided human image generation model using masked diffusion transformers, which improves scalability and efficiency over existing approaches.
https://openreview.net/forum?id=jEWpcEyuUl
Compressor summary: The paper reassesses the role of marginal confounder distribution in partial identification and provides a criterion for determining its impact on causal inference.
https://openreview.net/forum?id=j6rG1ETRyu
Compressor summary: The study proposes and tests algorithms for learning multiple distinct solutions from a single task in offline reinforcement learning.
https://openreview.net/forum?id=j6QZy90B93
Compressor summary: The paper proposes a new method, Hybrid Neural Representations for Spherical data (HNeR-S), to better handle nonlinear spherical signals like weather and CMB data using positional features from feature-grids and a multi-layer perceptron.
https://openreview.net/forum?id=j5wf1NNhFs
Compressor summary: The U-ext-Hop model improves memory retrieval in Hopfield networks by using a learnable feature map that transforms the energy function into kernel space, enhancing memory capacity and reducing confusion.
https://openreview.net/forum?id=j5csKrtyAe
Compressor summary: The text discusses the limitations of current methods to prevent misuse of large language models by censoring their outputs and calls for re-evaluation and new approaches to ensure safety.
https://openreview.net/forum?id=j56JAd29uH
Compressor summary: FADAS is a novel method that combines adaptive federated optimization with asynchronous updates, improving efficiency and resilience in privacy-preserving machine learning.
https://openreview.net/forum?id=j4HtfTqr0f
Compressor summary: The paper introduces MILP-FBGen, a machine learning technique for generating realistic and feasible Linear Programming and Mixed-Integer Linear Programming instances that preserve key properties and improve downstream task performance.
https://openreview.net/forum?id=j35VcooKG8
Compressor summary: The paper studies a hard Multi-Armed Bandit problem where neglected arms leave the game, proposes FC-SE algorithm with regret bounds, and extends it to handle new arms with FC-Entry algorithm.
https://openreview.net/forum?id=j2pLfsBm4J
Compressor summary: A new algorithmic framework for distributional reinforcement learning uses mean embeddings of return distributions and simple linear-algebraic computations to update the sketch Bellman operator, with theoretical and empirical evidence for its effectiveness.
https://openreview.net/forum?id=ixdfvnO0uy
Compressor summary: The article develops a general framework for optimizing vector-valued functions using multiparameter homological descriptors from persistent homology, showing that it improves performance over one-parameter descriptors.
https://openreview.net/forum?id=iup9NElHji
Compressor summary: The paper proposes a new transformer block that reduces the computational cost of visual transformers while maintaining or improving their accuracy, and introduces a novel variational upper bound for information bottleneck optimization.
https://openreview.net/forum?id=itYGbe0Cs1
Compressor summary: The paper introduces DR-MDPs to model preference changes in AI alignment, shows that static preferences can lead to undesirable AI influence, and explores potential solutions while acknowledging their limitations.
https://openreview.net/forum?id=itDhUBY2xf
Compressor summary: The paper proposes methods to train networks under partial differential equations, which require many evaluations, by addressing the bias caused by naive integral approximations.
https://openreview.net/forum?id=isUSVgS7W1
Compressor summary: Event cameras can reconstruct 3D scenes from sparse event data using a generalizable framework that estimates depth, intensity, and Gaussian regression, outperforming existing methods.
https://openreview.net/forum?id=iroZNDxFJZ
Compressor summary: The paper argues that large language models can revolutionize time series analysis, enabling efficient decision-making and new possibilities like modality switching and question answering.
https://openreview.net/forum?id=irBHPlknxP
Compressor summary: The paper proposes a new image restoration method using optimal transport and residual information to preserve the original image structure better than existing methods.
https://openreview.net/forum?id=iqAyWVLUEO
Compressor summary: This paper analyzes the theoretical properties of the Robust Satisficing (RS) model, a streamlined approach to robust optimization with better statistical guarantees and performance than existing methods.
https://openreview.net/forum?id=io1XSRtcO8
Compressor summary: PATH-WL is a powerful new method for graph neural networks that uses paths and shortest path distance information to achieve strong empirical results and solve complex problems like strongly regular graphs.
https://openreview.net/forum?id=inEuvSg0y1
Compressor summary: Mol-AE is a new auto-encoder model for 3D molecular representation learning with positional encoding and 3D Cloze Test objective that outperforms existing methods.
https://openreview.net/forum?id=ihv6pWuILN
Compressor summary: The study proposes a new causal model to analyze customer churn using tensor completion methods and shows its effectiveness in practice.
https://openreview.net/forum?id=igRjCCAz2a
Compressor summary: EDDPMs are versatile probabilistic models that combine encoding-decoding with diffusion for broad applicability and enhanced performance across text, proteins, and images.
https://openreview.net/forum?id=igRAPavrrS
Compressor summary: The paper analyzes standard differentially private gradient descent for linear regression, showing its accuracy and sample complexity match non-private methods with proper hyperparameter choices and adaptive confidence intervals.
https://openreview.net/forum?id=ie3vXkMvRY
Compressor summary: The text discusses how Reinforcement Learning (RL) improves session-based recommendation by promoting better embeddings of user interactions, and suggests using an auxiliary loss instead of RL to achieve similar performance gains.
https://openreview.net/forum?id=idyUNsoZ75
Compressor summary: SkewSize is a new metric that measures and characterizes bias in model predictions across subgroups, improving upon existing benchmarks.
https://openreview.net/forum?id=icijMMWwdG
Compressor summary: The paper studies smoothed online quadratic optimization with different costs and analyzes optimal algorithms for both adversarial and stochastic settings, proposing a new distribution-agnostic dynamic interpolation algorithm called Lazy Adaptive Interpolation (LAI).
https://openreview.net/forum?id=ibwxzYCep9
Compressor summary: The paper proposes a new framework for image restoration that balances visual quality and faithfulness to the original image using a stochastic degradation model and early-stopping.
https://openreview.net/forum?id=iaV2fU6Dif
Compressor summary: Llip is a new pretraining method for vision-language models that leverages diverse captions to improve image representation and performance on zero-shot tasks.
https://openreview.net/forum?id=ia5XvxFUJT
Compressor summary: The paper presents FlashLinearAttention and gated linear attention (GLA), which improve linear attention's efficiency and performance for parallel training and inference in Transformers.
https://openreview.net/forum?id=ia0Z8d1DbY
Compressor summary: PRODIGY is a method to generate graphs with precise control using pre-trained diffusion models, handling both soft and hard constraints, achieving high constraint satisfaction for various applications like drug discovery.
https://openreview.net/forum?id=iYYA5zDoCm
Compressor summary: This paper introduces a decentralized model for multi-agent systems with varying dependencies and proposes three near-optimal, scalable policies for it.
https://openreview.net/forum?id=iUwHnoENnl
Compressor summary: The paper proposes KTO, a new human-aware loss function for LLMs that directly maximizes human utility based on Kahneman-Tversky prospect theory, and shows its effectiveness compared to other HALOs and cross-entropy minimization at different scales.
https://openreview.net/forum?id=iRcmqXZjeK
Compressor summary: DML-IV is a non-linear IV regression method that reduces bias in two-stage IV regressions and effectively learns high-performing policies using a novel learning objective and DML framework.
https://openreview.net/forum?id=iQTElQbAqo
Compressor summary: The text proposes a new method to compare differentially private machine learning mechanisms by measuring their worst-case privacy risks and shows its usefulness through examples.
https://openreview.net/forum?id=iPFuWc1TV2
Compressor summary: The Triplet Graph Transformer is a novel model that uses triplet attention and aggregation to capture third-order interactions in graphs, enabling better molecular property prediction and achieving state-of-the-art results on various benchmarks.
https://openreview.net/forum?id=iOEReiiTit
Compressor summary: The paper proposes a novel certification method for machine learning models that uses a multi-level hierarchy to reduce abstain rates and increase information gain compared to existing methods.
https://openreview.net/forum?id=iLyUEPZ0fR
Compressor summary: Block coordinate descent with optimal stepsizes can achieve faster convergence than gradient descent and momentum for some problems.
https://openreview.net/forum?id=iLfk2CwEHA
Compressor summary: DeepPolar codes use neural networks to generalize and improve Polar codes for error correction at short-to-medium block lengths.
https://openreview.net/forum?id=iLSgF7jMtI
Compressor summary: CogDPM is a new model that integrates diffusion probabilistic models with predictive coding theory to improve visual world prediction by incorporating precision weighting mechanism.
https://openreview.net/forum?id=iLCZtl7FTa
Compressor summary: Debate can help weaker AI models assess stronger ones without human labels, and optimizing expert debaters for persuasiveness improves their performance.
https://openreview.net/forum?id=iKkFruh4d5
Compressor summary: This paper studies how multi-pass gradient descent with batch re-use improves learning for two-layer neural networks when target functions have multiple indexes.
https://openreview.net/forum?id=iJlPJsTw2B
Compressor summary: The paper proposes a new optimal format for large language models and explores the tradeoff between accuracy and chip area using different datatypes.
https://openreview.net/forum?id=iJWeK2snMH
Compressor summary: The paper proposes a method to calibrate probabilistic forecasts for multi-dimensional outputs using highest density regions, which considers the joint distribution across dimensions.
https://openreview.net/forum?id=iHSgfGob9j
Compressor summary: CoLoRA is a method that uses low-rank adaptive neural networks to predict the evolution of solution fields for partial differential equations, achieving fast and accurate results even in data-scarce regimes.
https://openreview.net/forum?id=iGMTxygzcJ
Compressor summary: A quantum algorithm using quantum Fourier transforms can efficiently estimate Gibbs measures for periodic functions and improve sampling precision in high temperature regimes.
https://openreview.net/forum?id=iE2lMjeXRR
Compressor summary: The paper explores the theoretical advantages of Adam optimizer in optimization problems using an online learning framework.
https://openreview.net/forum?id=iC8l9DI1ZX
Compressor summary: Supervised contrastive representation learning methods, SupSiam and SupBYOL, improve on self-supervised learning by using labels to reduce intra-class variance and avoid collapse, leading to better results across various tasks and datasets.
https://openreview.net/forum?id=i9C4Kwm56G
Compressor summary: The study proposes a novel task for machine learning models that mimics animals' ability to adapt quickly and maintain proficiency, using a combination of neural networks and biological inspirations.
https://openreview.net/forum?id=i56plqPpEa
Compressor summary: The authors present a theory for how simple language models can perform complex tasks by predicting the next token in a sequence and show that linear networks and MLPs have surprising abilities in text generation and arithmetic tasks.
https://openreview.net/forum?id=i0nVanexij
Compressor summary: The paper proposes self-correction functions to stabilize generative models trained on synthetic data, especially for tasks like human motion synthesis.
https://openreview.net/forum?id=hz8cFsdz7P
Compressor summary: The Scientific Generative Agent (SGA) combines the strengths of large language models and simulations to enhance scientific discovery by proposing hypotheses, reasoning about discrete components, and receiving feedback on continuous parts.
https://openreview.net/forum?id=hunSEjeCPE
Compressor summary: The paper introduces EDIS, a method that uses a diffusion model to extract prior knowledge from offline data and energy functions to distill it for better online learning, improving efficiency and safety in RL.
https://openreview.net/forum?id=htq0FbPOsY
Compressor summary: The article studies the computational complexity of the SHAP framework for local explainability of ML models under realistic assumptions and shows that it can be computed in polynomial time for some model families.
https://openreview.net/forum?id=hsHIxrnrMx
Compressor summary: The paper proposes a method to match text representations in tasks with different domains by optimizing information bottleneck, which improves the performance of such tasks.
https://openreview.net/forum?id=hrwIndai8e
Compressor summary: The paper introduces P2L, a method that uses text-to-image latent diffusion models with prompt tuning to solve imaging inverse problems better than existing methods.
https://openreview.net/forum?id=hrWte3nlzr
Compressor summary: The paper proposes a new primal-dual algorithm for constrained reinforcement learning that avoids error cancellations and achieves sublinear regret.
https://openreview.net/forum?id=hr8OXXMb7a
Compressor summary: This paper introduces StoP, a method that incorporates location uncertainty in MIM by using stochastic positional embeddings, which improves downstream performance on various tasks.
https://openreview.net/forum?id=hqNz4LDuhn
Compressor summary: The text introduces a new score function estimator for diffusion generative models that reduces variance and improves training speed and sample quality.
https://openreview.net/forum?id=hoVwecMqV5
Compressor summary: VQ-BeT is a new model that improves on BeT for generating complex behaviors by tokenizing continuous actions with hierarchical vector quantization, achieving faster inference speed and better performance in various environments.
https://openreview.net/forum?id=hnqlgwcRxb
Compressor summary: The paper investigates theoretical properties of variational inference for non-Gaussian Mixture of Gaussians, showing how it can be cast as optimizing Dirac positions using gradient descent and studying errors in the process.
https://openreview.net/forum?id=hlvKd7Vdxm
Compressor summary: The paper proposes ExCP, a framework that compresses large language models' checkpoints significantly while maintaining high accuracy on various tasks.
https://openreview.net/forum?id=hgHQvrvwH9
Compressor summary: The paper proposes an easy-to-use recipe to improve privacy profiles of ReportNoisyMax and PrivateTuning using base algorithms' privacy profiles, leading to better private learning experiments.
https://openreview.net/forum?id=hg4wXlrQCV
Compressor summary: The paper presents Craftax-Classic, a faster version of Crafter, and Craftax, a more challenging benchmark for RL research that requires deep exploration and planning.
https://openreview.net/forum?id=hdpv6mall8
Compressor summary: The text discusses the potential harmful applications of machine learning and the need for evaluating the European Union's Digital Services Act to curb these harms.
https://openreview.net/forum?id=hcQfTsVnBo
Compressor summary: The text describes how researchers reverse engineered a neural network that learned the arithmetic of permutation groups and challenges the interpretability of another paper on the same topic.
https://openreview.net/forum?id=hcASxFvmZ5
Compressor summary: PEAK is a novel nonparametric sequential test that uses a betting scheme to control type-I error and power in multiple data streams, reducing sample complexity and computational overhead.
https://openreview.net/forum?id=hbsKxUEreL
Compressor summary: The paper proposes an algorithm for anomaly detection that adapts to distribution shifts, has low false positive and negative rates, and uses offline data effectively.
https://openreview.net/forum?id=haUOhXo70o
Compressor summary: SMT is a novel multi-task RL algorithm that prioritizes harder tasks, uses a task difficulty metric for efficient resource allocation, and resets network parameters to mitigate simplicity bias.
https://openreview.net/forum?id=hZ0fWhgVch
Compressor summary: The paper proposes a training-free approach to improve text-to-image alignment by optimizing images directly with the supervision of vision-language models and incorporating score distillation sampling.
https://openreview.net/forum?id=hYHsrKDiX7
Compressor summary: GaLore is a training strategy for large language models that reduces memory usage by up to 82.5% without sacrificing performance, enabling pre-training on 7B models with consumer GPUs and 24GB memory.
https://openreview.net/forum?id=hXQOO6VsxH
Compressor summary: The paper proposes two efficient algorithms for RL with Aggregate Bandit Feedback, which allows feedback at episode end instead of individual rewards, and achieves near-optimal regret guarantees using linear function approximation and new randomization techniques.
https://openreview.net/forum?id=hWng0GXeE4
Compressor summary: The paper presents a novel geometry-based coreset construction method that efficiently selects training data to reconstruct the decision boundary of a deep neural network, achieving high data pruning rate with minimal accuracy loss and showing strong cross-architecture transferability.
https://openreview.net/forum?id=hTiNFCNxM1
Compressor summary: The paper proposes DCEM, an algorithm to learn from selective labels with disparate censorship, and shows that it reduces bias without sacrificing performance on synthetic and real clinical data.
https://openreview.net/forum?id=hRX1o7FBhT
Compressor summary: The paper introduces Progressive Inference, a method to explain decoder-only transformer models' predictions by evaluating intermediate classifications at different input positions, using either Single Pass or Multi Pass approaches.
https://openreview.net/forum?id=hRBdOHVn7y
Compressor summary: The paper studies online metric problems with long-term constraints for resource allocation in sustainable energy/computing systems, and proposes optimal algorithms for bounded hitting cost gradients and weighted $\ell_1$ metrics.
https://openreview.net/forum?id=hQpUhySEJi
Compressor summary: The paper introduces Subequiriant Hierarchical Neural Networks (SHNN) for learning policies in multi-entity 3D environments, which use task assignment and subequivariance to reduce complexity and improve performance on a new benchmark called Multi-entity Benchmark (MEBEN).
https://openreview.net/forum?id=hLuNVjRnY3
Compressor summary: The authors propose a new model merging algorithm called CCA Merge, which uses Canonical Correlation Analysis to maximize the correlations between linear combinations of the models' features and improves performance over past methods in both two-model and multi-model settings.
https://openreview.net/forum?id=hLGxDYo0eF
Compressor summary: The paper analyzes the performance of a policy optimization reinforcement learning algorithm that learns from human feedback without knowing the reward function, providing bounds on query complexity and novel techniques for inferring reward parameters.
https://openreview.net/forum?id=hKdJPMQvew
Compressor summary: HALO is a hyperbolic neural network that uses epistemic uncertainty to select data points for pixel-level active learning, achieving state-of-the-art results in semantic segmentation under domain shift with minimal supervision.
https://openreview.net/forum?id=hJaWoU3Emh
Compressor summary: The paper proposes a two-stage method to improve machine learning models' performance and generalization when data distributions vary across multiple segments of the population, using linear combinations and refinement steps for each segment.
https://openreview.net/forum?id=hG6gddAKnJ
Compressor summary: The paper explores and characterizes conservation laws in non-Euclidean geometries and momentum-based dynamics for neural network training.
https://openreview.net/forum?id=hFEgae0od4
Compressor summary: The paper proposes a deep learning method that selects synergistic feature subsets from multi-view data using interaction information to understand target outcomes better.
https://openreview.net/forum?id=h8aTi32tul
Compressor summary: SiBBlInGS is a graph-based method for discovering interpretable units in time series data across different states that captures complex variability and adapts to varying session lengths and sample sizes.
https://openreview.net/forum?id=h3SGdpI4Ta
Compressor summary: The paper studies how to control linear systems with uncertain disturbances within a Wasserstein-2 ambiguity set, and proposes efficient algorithms to compute near-optimal control policies.
https://openreview.net/forum?id=h2uBuQvpp8
Compressor summary: ASMR is a method that learns to selectively sample k-space measurements for faster and accurate disease detection in MR imaging.
https://openreview.net/forum?id=gzis9n5r7e
Compressor summary: The paper proposes a method called Transition Discriminator-based Imitation Learning (TDIL) that uses a transition discriminator to compute surrogate rewards from one expert trajectory, addressing reward sparsity and achieving expert-level performance in single-demonstration imitation learning.
https://openreview.net/forum?id=gxOQEMRbRa
Compressor summary: Q-probing adapts pre-trained language models to new tasks using a linear function on embeddings that reweights candidate completions based on task-specific rewards or policy objectives, and can improve performance in data-limited regimes.
https://openreview.net/forum?id=guFsTBXsov
Compressor summary: Minimal Frame Averaging (MFA) is a new framework for achieving exact equivariance in machine learning systems by constructing minimal frames that encode symmetries efficiently and effectively across various tasks, including physics simulations and complex-valued domains.
https://openreview.net/forum?id=gu3nacA9AH
Compressor summary: Generalized preference optimization is a unified framework for offline learning that encompasses existing methods and allows tuning of regularization through convex functions.
https://openreview.net/forum?id=gtYdvSGMYV
Compressor summary: LAGMA improves cooperative multi-agent reinforcement learning by generating a goal-reaching trajectory in latent space and providing an incentive for agents to follow it.
https://openreview.net/forum?id=gqA8ZHO0j8
Compressor summary: USBS is a fast and scalable spectral bundle method for solving SDPs that can leverage warm-start initialization, achieving dramatic speedups compared to existing methods.
https://openreview.net/forum?id=gn5AsHIIwb
Compressor summary: StackSight is a novel technique that combines LLMs with program analysis to decompile complex WebAssembly code into readable C++ snippets, improving understanding and decompilation.
https://openreview.net/forum?id=glfcwSsks8
Compressor summary: The authors investigate the geometry of Large Language Models (LLMs) to understand their inner mechanisms and develop novel solutions for tasks like toxicity detection.
https://openreview.net/forum?id=gjoUXwuZdy
Compressor summary: The authors introduce VisionGraph, a benchmark for testing large multimodal models' ability to solve graph theory problems using a Description-Program-Reasoning chain that improves their accuracy and performance.
https://openreview.net/forum?id=gjgRKbdYR7
Compressor summary: SelfIE is a framework that lets large language models explain their own reasoning in natural language, enabling control over their responses for reliability, transparency, and future development.
https://openreview.net/forum?id=girxGkdECL
Compressor summary: Key points: - LLMs can make factual errors in open-domain question answering - The paper explores if AI assistants can know what they don't know and express it through natural language - They create an Idk dataset and align the assistant with it using different methods - The aligned assistant is more truthful and declines less often Summary: The paper investigates how to make AI assistants aware of their knowledge gaps and express them in natural language, creating a new dataset and alignment methods that improve their truthfulness.
https://openreview.net/forum?id=ghYrfdJfjK
Compressor summary: PolySketchFormer is a fast and accurate Transformer-based language model that replaces softmax attention with polynomial attention and uses sketching techniques for linear-time computation.
https://openreview.net/forum?id=ghNRg2mEgN
Compressor summary: The study examines whether weak model supervision can elicit the full capabilities of stronger models and finds that while there is some success, techniques like reinforcement learning may be needed for superhuman models without further work.
https://openreview.net/forum?id=geajNKab7g
Compressor summary: FOMA is a new domain-independent data augmentation method for regression problems that samples from the tangent planes of the train distribution and improves generalization and robustness.
https://openreview.net/forum?id=gbD9MAc9p0
Compressor summary: The paper proposes quality-weighted Vendi scores to balance quality and diversity in experimental design, improving data collection and discovery in various applications.
https://openreview.net/forum?id=gWEwIlZrbQ
Compressor summary: The paper proposes a novel DME method for federated learning that improves the NMSE guarantee and computational efficiency by using off-the-shelf solvers and quantization.
https://openreview.net/forum?id=gVjMwLDFoQ
Compressor summary: iDEM is a fast and scalable algorithm that uses only the energy function and its gradient to generate independent samples from unnormalized probability distributions, achieving state-of-the-art performance on various tasks.
https://openreview.net/forum?id=gVg8V9isul
Compressor summary: CTAN is a new method for modeling spatio-temporal information in C-TDGs that outperforms existing methods on long-range tasks.
https://openreview.net/forum?id=gUFufRkzjV
Compressor summary: The text proposes a framework to generate Verification-Friendly Neural Networks that balance prediction performance and verification-friendliness, enabling more robustness in safety-critical applications.
https://openreview.net/forum?id=gTBjkJvadC
Compressor summary: Large batch sizes reduce gradient variance in differentially private stochastic gradient descent (DP-SGD) by decreasing subsampling-induced variance and are beneficial especially in the asymptotic regime.
https://openreview.net/forum?id=gSMUjrkRRk
Compressor summary: Quasi-Monte Carlo methods can improve kernel approximation error for certain kernels, leading to fewer random features needed and better performance in kernel ridge regression.
https://openreview.net/forum?id=gS3nc9iUrH
Compressor summary: The paper proposes a data-efficient and interpretable graph grammar model for representing and reasoning over complex molecular structures, enabling better design and property prediction.
https://openreview.net/forum?id=gQz30hTkRE
Compressor summary: The paper proposes machine learning methods to understand the group structure of homotopy groups of spheres by generating simplicial cycles from algorithmic datasets related to Dyck languages.
https://openreview.net/forum?id=gQpBnRHwxM
Compressor summary: The text discusses the importance of designing AI systems to serve diverse human values and suggests a roadmap for achieving pluralistic alignment using large language models and different types of benchmarks.
https://openreview.net/forum?id=gPStP3FSY9
Compressor summary: The paper introduces Cross-Risk Minimization (XRM), an algorithm that automatically discovers environments within datasets for robust out-of-distribution generalization without relying on human-annotated environment labels.
https://openreview.net/forum?id=gPBMkJG7bt
Compressor summary: The paper studies how test risk changes with learning rate in continuous time stochastic gradient flow dynamics and applies the theory to a weak features model, finding the impact of stochasticity on test risk.
https://openreview.net/forum?id=gL5djEYLx2
Compressor summary: This paper improves existing bounds on the maximum diameter and cohesion of complete-link and average-link hierarchical clustering algorithms for metric spaces, supporting the preference of complete-link over single-link for producing compact clusters.
https://openreview.net/forum?id=gKPkipJ3gm
Compressor summary: This paper introduces Causal-IQA, an end-to-end blind Image Quality Assessment method that uses causality to improve estimation accuracy and generalization by mitigating confounding effects between distortion types, image contents, and human ratings.
https://openreview.net/forum?id=gEbl6XNLK6
Compressor summary: CB2Ms are memory-enhanced concept bottleneck models that learn from past interventions and can generalize them to new situations, improving interpretability and performance.
https://openreview.net/forum?id=gE7qZurGH3
Compressor summary: The paper proposes a new method for lossless graph condensation by using curriculum learning, expanding window matching, and a customized loss function to train expert trajectories that preserve the original graph's structure and performance.
https://openreview.net/forum?id=gDQuupz8mm
Compressor summary: This paper proposes new algorithms that achieve small-loss adaptive regret bounds for various types of convex functions and handle changing environments.
https://openreview.net/forum?id=gB3E8IwQZy
Compressor summary: Key points: - The paper introduces GSGD, a novel gradient sketching method for large-scale optimization - GSGD does not require importance sampling but can match the convergence rate of methods with it - GSGD can exploit non-smooth regularization terms for faster convergence - Experimental results show the effectiveness and efficiency of GSGD Summary: The paper presents GSGD, a new gradient sketching method that achieves fast convergence in large-scale optimization without importance sampling or exploiting smooth regularization terms.
https://openreview.net/forum?id=gAyzjHw2ml
Compressor summary: SceneCraft is an LLM agent that creates Blender scripts from text descriptions to render complex scenes, using spatial planning, image analysis, and library learning.
https://openreview.net/forum?id=g9mYBdooPA
Compressor summary: Policy-conditioned model (PCM) learning improves reinforcement learning by adapting the dynamics model to different evaluation policies for better prediction accuracy.
https://openreview.net/forum?id=g8AigOTNXL
Compressor summary: The paper proposes SCG, a new method for symbolic music generation that works with non-differentiable rules and improves quality and controllability over existing methods.
https://openreview.net/forum?id=g89jAdrnAF
Compressor summary: Key points: - The paper proposes a method to predict the effects of amino acid mutations on protein-protein binding using hierarchical prompt learning - The method models the joint distribution of each mutation with various microenvironmental features - The method outperforms existing pre-training-based methods and can be applied to optimize antibodies against SARS-CoV-2 Summary: The paper introduces a hierarchical prompt learning framework that predicts how amino acid mutations affect protein-protein binding by modeling the microenvironmental changes. The method is more efficient and effective than previous pre-training methods and can help design better antibodies against COVID-19.
https://openreview.net/forum?id=g43yUNWX4V
Compressor summary: The paper proposes two algorithms for Federated Reinforcement Learning (FRL) that can handle large levels of environment heterogeneity and achieve state-of-the-art convergence results.
https://openreview.net/forum?id=g1Gf0hoPSz
Compressor summary: The paper proposes a latent-variable model that uses Gaussian processes to handle high-dimensional data with missing values and unknown measurement times in various fields.
https://openreview.net/forum?id=fz9PaJNViP
Compressor summary: MOKD is a bi-level optimization framework that improves few-shot classification by learning class-specific representations and maximizing kernel dependence between them and labels, while minimizing dependence among all samples.
https://openreview.net/forum?id=fywWm06IGn
Compressor summary: The paper proposes a method to identify features that have different importance in subgroups of a dataset, which can help detect and rectify biases in classifiers trained on such data.
https://openreview.net/forum?id=fwxnHViGNj
Compressor summary: The paper compares multi-task benchmarks in machine learning to electoral systems, shows a trade-off between diversity and sensitivity to irrelevant changes, and introduces new quantitative measures and approximation algorithms for these aspects.
https://openreview.net/forum?id=fuX4hyLPmO
Compressor summary: The paper introduces SLEB, a new pruning method that removes redundant transformer blocks from large language models to speed up inference while maintaining accuracy and perplexity.
https://openreview.net/forum?id=ft5jK9uPgC
Compressor summary: No algorithm beats uniform sampling in A/B testing with fixed budget and Bernoulli rewards.
https://openreview.net/forum?id=fsVBsxjRER
Compressor summary: The authors propose LocoGen and LocoEdit, methods that use causal tracing to locate and edit specific visual attributes in text-to-image models, enabling efficient model editing.
https://openreview.net/forum?id=frA0NNBS1n
Compressor summary: The paper proposes using Sequential Monte Carlo with learned twist functions to improve probabilistic inference for various language model techniques such as safety, capability, and harmlessness training.
https://openreview.net/forum?id=fqeANcjBMT
Compressor summary: DP-BiTFiT is a new method for privacy-preserving fine-tuning of large pre-trained models that achieves high accuracy, efficiency, and parameter efficiency.
https://openreview.net/forum?id=fqPH6ejwGi
Compressor summary: This paper explores different types of neural network architectures encodings, introduces unified encodings, and proposes FLAN, a predictor that significantly reduces the cost of training NAS accuracy predictors.
https://openreview.net/forum?id=fq0NaiU8Ex
Compressor summary: DARE is a method that sparsifies and merges language model parameters from different tasks without retraining, enabling larger models with diverse capabilities and improved performance.
https://openreview.net/forum?id=fowZNENcVJ
Compressor summary: The text discusses how providing uncertainty information from AI can enhance human decision-making and reports positive results from two experiments testing this idea.
https://openreview.net/forum?id=foPMkomvk1
Compressor summary: The text proposes a method to achieve client-level fairness in federated learning by using an adaptive aggregation scheme that optimizes online convex optimization and improves decision making for cross-device and cross-silo settings.
https://openreview.net/forum?id=fiugPLSXjK
Compressor summary: The paper proposes a robust and efficient incomplete multi-view clustering method called EDISON, which uses an enhanced dictionary representation strategy and Gaussian error rank approximation for tensor data.
https://openreview.net/forum?id=fgBWtOw66T
Compressor summary: SFC is a new transform for fast and accurate quantized convolution using symbolic computing and the Discrete Fourier Transform, achieving better efficiency than Winograd and FFT algorithms.
https://openreview.net/forum?id=ffS0aYP6mk
Compressor summary: The paper studies how frequently communicating in Federated Learning affects generalization error and provides bounds and experiments for different learning models.
https://openreview.net/forum?id=ffLblkoCw8
Compressor summary: MAGDi is a method to improve reasoning in smaller models by distilling knowledge from multiple LLMs using graph representations and three objective functions, achieving efficiency and generalization improvements.
https://openreview.net/forum?id=fdroxYsgzQ
Compressor summary: Summary: The paper proposes Prompting for Robustness (PfR), a method to improve machine learning models' robustness against spurious correlations by using foundation models to predict the spurious attribute and learn a classifier that performs well across different labels.
https://openreview.net/forum?id=fVg9YrSllr
Compressor summary: The text introduces a benchmark to evaluate sampling methods from intractable distributions using a standardized task suite and various performance criteria, including new metrics for mode collapse.
https://openreview.net/forum?id=fSnMqHZ8xr
Compressor summary: BE-CBO is a new Bayesian optimization method that learns constraints with neural networks to efficiently explore the boundary between feasible and infeasible designs in black-box optimization problems.
https://openreview.net/forum?id=fSNHK7mu3j
Compressor summary: The paper shows that Graph Neural Networks (GNNs) often use the input graph structure even when it's not needed, leading to suboptimal solutions, and suggests using regular graphs to improve performance and avoid this bias.
https://openreview.net/forum?id=fRG45xL1WT
Compressor summary: FeatLLM is a novel in-context learning framework that uses LLMs to generate optimal features for tabular predictions, achieving high performance few-shot learning.
https://openreview.net/forum?id=fPwWfoyxL1
Compressor summary: Key points: - The paper proposes a zeroth-order algorithm for optimization problems on Riemannian manifolds with improved efficiency and robustness. - The algorithm achieves state-of-the-art function query complexity and almost sure convergence in the asymptotic sense. - The algorithm requires larger smoothing parameters, improving the existing result by a factor of $ ilde{\mathcal{O}}(\epsilon^{7/8}d^{-1/2})$. Summary: The paper presents a Riemannian accelerated zeroth-order algorithm that is more efficient and stable than previous ones for optimization problems on Riemannian manifolds.
https://openreview.net/forum?id=fOBas5H4Xc
Compressor summary: Key points: - The paper proposes an algorithm to learn low-dimensional models from high-dimensional observations of LTI systems. - The algorithm has an optimal sample complexity up to logarithmic factors and dimension-independent constants. - The paper also considers a meta-learning problem where the observer column space can be learned from multiple LTI systems. Summary: The paper presents an optimal algorithm for learning low-dimensional models of LTI systems from high-dimensional data, and extends it to a meta-learning problem with collective observer column space learning.
https://openreview.net/forum?id=fO31YAyNbI
Compressor summary: The paper proposes MotionEpic, a novel video Multimodal Large Language Model, and VoT, a Video-of-Thought reasoning framework that use spatial-temporal scene graphs and Chain-of-Thought techniques to improve video understanding and reasoning.
https://openreview.net/forum?id=fNJbcxhxRj
Compressor summary: The paper presents MorseDet, a novel computer vision method that uses algebraic topology tools to learn keypoints with scale-invariant and flexible performance.
https://openreview.net/forum?id=fM9xTkpAdu
Compressor summary: RAOQ is a method to improve in-memory computing by mitigating ADC quantization error and adapting AI models for better performance in computer vision and NLP tasks.
https://openreview.net/forum?id=f8G2KSCSdp
Compressor summary: The paper introduces CoOPood, a fine-grained prompt tuning method for VLMs that aligns text with invariant features and avoids spurious ones, improving OOD generalization.
https://openreview.net/forum?id=f6QenZyyeP
Compressor summary: The paper proposes two methods to optimize Neural ODEs, improving their training and inference speed by drawing inspiration from control theory.
https://openreview.net/forum?id=f5gtX2VWSB
Compressor summary: A growable and modular neural network architecture can learn continually from previous tasks without forgetting or interference, and scales well with the number of tasks.
https://openreview.net/forum?id=f49AkFT5jf
Compressor summary: The paper proposes new black-box data poisoning attacks against conformal prediction methods, which can manipulate the uncertainty of specific examples more effectively than traditional attacks.
https://openreview.net/forum?id=f47ZK6gy3I
Compressor summary: The paper introduces FineSSL, a new semi-supervised learning approach that adapts pre-trained foundation models, improving their performance, reducing training cost, and integrating with other algorithms.
https://openreview.net/forum?id=f3TUipYU3U
Compressor summary: HarmBench is a standardized framework for evaluating automated red teaming methods against large language models, revealing new insights and improving LLM robustness.
https://openreview.net/forum?id=eyxVRMrZ4m
Compressor summary: The authors propose a method to improve reinforcement learning for language models by using attention weights from the reward model to redistribute the reward, making it easier to optimize and potentially leading to better results.
https://openreview.net/forum?id=eqY64Z1rsT
Compressor summary: FILM is a novel image fusion method that uses textual descriptions generated by ChatGPT to guide the fusion process, enhancing feature extraction and contextual understanding.
https://openreview.net/forum?id=eqIGoEoI10
Compressor summary: The paper proposes adaptive policies for A/B testing that estimate the average treatment effect with a desired confidence interval width and probability, using an optimal sample size lower bound derived from a non-convex optimization problem.
https://openreview.net/forum?id=eo88noTbb5
Compressor summary: The paper analyzes how agnostic learning of mixed linear regression can be achieved by EM and AM algorithms without assuming generative models.
https://openreview.net/forum?id=emtXYlBrNF
Compressor summary: The paper proposes a method to attribute and understand large language models by extending Layer-wise Relevance Propagation to handle attention layers, improving faithfulness and efficiency over existing methods.
https://openreview.net/forum?id=elF0QoBSFV
Compressor summary: NDOT is a new online training method for SNNs that uses neuronal dynamics to compute gradients efficiently and accurately on large-scale datasets.
https://openreview.net/forum?id=elCOPIm4Xw
Compressor summary: CLOPPS is a machine learning model that predicts clinical outcomes using information from different data sources and performs better than existing models in real-world scenarios.
https://openreview.net/forum?id=efzkSbpyRw
Compressor summary: The paper studies how split Conformal Prediction performs on Markovian data, showing its coverage gap depends on the mixing time of the chain, and proposes a method called $K$-split CP that adapts to the data's properties.
https://openreview.net/forum?id=eejhD9FCP3
Compressor summary: The IRDiff model uses a network of protein-molecule interactions to generate ligands that bind well to specific proteins, using references with desired properties as guidance.
https://openreview.net/forum?id=edHLN40DWu
Compressor summary: Mixture-of-Prompts (MoP) is a method that automates instruction design for large language models by dividing the problem space into sub-regions, each governed by a specialized expert with an instruction and demos, achieving high win rates on benchmarks.
https://openreview.net/forum?id=ecvuJWE1YY
Compressor summary: The text discusses using supervised data subset selection and active learning techniques to reduce data acquisition costs for machine learning models, but considering the impact of future data deletions under GDPR and proposing deletion-anticipative data selection methods to optimize utility.
https://openreview.net/forum?id=ecnpYYHjt9
Compressor summary: DiffS4L uses diffusion models to generate synthetic speech data with different variations, improving self-supervised learning for low-resource languages and under privacy concerns.
https://openreview.net/forum?id=ecO7WOIlMD
Compressor summary: MF-CLR is a self-supervised contrastive learning method for representing multi-frequency time series data, achieving excellent results on various financial downstream tasks.
https://openreview.net/forum?id=ebt5BfRHcW
Compressor summary: The paper presents a new mixture-of-expert approach, DirMixE, that captures both global and local variations in test label distributions for long-tail recognition tasks.
https://openreview.net/forum?id=eapFRURALQ
Compressor summary: The PHE-trinity model explores how the hippocampus contributes to language comprehension by using a modular continuous attractor network to represent syntactic structure and two separate input streams, and demonstrates its effectiveness in learning from limited data.
https://openreview.net/forum?id=eaNLvrP8n1
Compressor summary: AVTrack is a framework that adapts the computation of transformer-based visual trackers for real-time UAV tracking by dynamically optimizing ViT architecture and learning view-invariant representations.
https://openreview.net/forum?id=ea2MgKn3sV
Compressor summary: The paper explores how large language models can revolutionize black-box optimization by using their comprehension, flexibility, and performance prediction abilities.
https://openreview.net/forum?id=eZiQWM5U0E
Compressor summary: The paper proposes a new method (HJFBiO) for solving nonconvex-PL bilevel optimization problems without Hessian/Jacobian matrices, with optimal convergence and gradient complexity.
https://openreview.net/forum?id=eY98MVffrD
Compressor summary: The paper proposes learning-rate-free algorithms for stochastic optimization over Riemannian manifolds that eliminate hand-tuning and provide optimal convergence guarantees.
https://openreview.net/forum?id=eY4jrFe6Qc
Compressor summary: The paper analyzes how kernel ridge regression with random feature mapping performs on large-scale nonparametric regression with different types of data dependence structures, showing optimality under exponential decay but sub-optimality under polynomial decay.
https://openreview.net/forum?id=eW0pZmziBH
Compressor summary: The paper introduces a fast and accurate statistical algorithm for inference under the Partial Credit Model used in psychometrics and other fields, with applications to education, recommendation systems, and finance.
https://openreview.net/forum?id=eVlx8DaG9h
Compressor summary: StrokeNUWA is a method that uses vector graphics and semantic "stroke" tokens to enable more natural and efficient visual synthesis with large language models.
https://openreview.net/forum?id=eVGpdivOnQ
Compressor summary: The study investigates if large language models can plan asynchronously, finding that they struggle without illustrations, and proposes a new technique called Plan Like a Graph to improve performance.
https://openreview.net/forum?id=eRThYD9BGD
Compressor summary: The paper proposes a new training method (PLP) that improves uncertainty calibration in deep neural networks by avoiding over-compression of top layers and using weak classifier heads.
https://openreview.net/forum?id=eQaOb4r6YC
Compressor summary: The paper proposes a fast and effective out-of-distribution detector that uses feature distances to decision boundaries without auxiliary models, achieving good performance and low latency.
https://openreview.net/forum?id=ePDnv4xESI
Compressor summary: MixPro is a data-efficient method for adapting pretrained models to new distributions using mixed embeddings and linear classifiers.
https://openreview.net/forum?id=eP3vsbB5wW
Compressor summary: The paper proposes a method to prune deep neural network ensembles under distribution shifts using a topology graph, improving predictive diversity and generalization performance on out-of-distribution data.
https://openreview.net/forum?id=eOtjMYdGLt
Compressor summary: Characteristic guidance is a non-linear correction method for DDPMs that improves semantic features and image quality by respecting the Fokker-Planck equation without additional training.
https://openreview.net/forum?id=eN1T7I7OpZ
Compressor summary: The paper presents Shūkai, a DRL agent for fighting games that improves generalizability and sample efficiency with Heterogeneous League Training and specific rewards, and shows its effectiveness in Naruto Mobile.
https://openreview.net/forum?id=eMQyb1tvvc
Compressor summary: The paper proposes a method to compress spiking neural networks by dynamically pruning and regenerating convolutional kernels based on their activity levels, achieving low-power and high-efficiency performance while maintaining model accuracy.
https://openreview.net/forum?id=eJFQROkaj0
Compressor summary: The paper introduces a new framework for robotic manipulation that combines multimodal perception and planning using tailored language models and retrieval-based policy learning.
https://openreview.net/forum?id=eGZH3HCuGm
Compressor summary: This paper investigates how simplicity bias affects general neural networks, especially two-layer ones, and suggests that features learned in the middle stages of training may improve out-of-distribution generalization.
https://openreview.net/forum?id=eG42XBhV9a
Compressor summary: The paper introduces OLLIE, a method that improves offline-to-online Imitation Learning by learning a better policy initialization and an aligned discriminator initialization, achieving better performance and efficiency in various domains.
https://openreview.net/forum?id=eFvoL7BOny
Compressor summary: The paper proposes a novel quantum reinforcement learning algorithm that achieves provably efficient exploration-exploitation trade-off and breaks the $\Omega(\sqrt{T})$-regret barrier in classical RL.
https://openreview.net/forum?id=eFSppFiVYG
Compressor summary: The paper proposes new generalization bounds for heavy-tailed stochastic optimization algorithms using fractional Fokker-Planck equation and shows that heavy tails can be beneficial or harmful depending on the problem structure.
https://openreview.net/forum?id=eDtty9ZCvt
Compressor summary: The paper proposes AutoActivator, a connectionist model with adaptive neural unit dynamics for class-incremental learning, which can expand its capacity when needed and reactivate required units at inference time without forgetting old classes.
https://openreview.net/forum?id=eDjvSFOkXw
Compressor summary: Lookahead decoding is a parallel algorithm that accelerates large language model decoding without needing auxiliary models or data stores, achieving up to 4x speedup on multiple GPUs.
https://openreview.net/forum?id=eCCaHZKdl4
Compressor summary: The Uncertainty-aware Reward Model (URM) improves instruction following in language models by estimating response quality and uncertainty using Bayesian approximation, leading to better performance on benchmarks.
https://openreview.net/forum?id=eC1OOpOGZW
Compressor summary: FORGrad is a new method that improves the performance of white-box attribution methods by filtering out high-frequency artifacts in gradient signals, making them more accurate and computationally efficient for model explanations.
https://openreview.net/forum?id=e93ffDcpH3
Compressor summary: The text explores a new language model architecture called BASED that balances memory efficiency and recall ability by combining linear and sliding window attention, achieving competitive results on perplexity and real-world tasks.
https://openreview.net/forum?id=e76GrGhIgf
Compressor summary: The paper explores the dense-to-sparse gating MoE, proposes a novel activation gate to improve convergence rates, and validates the results with simulations.
https://openreview.net/forum?id=e5tA3Apbmy
Compressor summary: The paper proposes a user response model to adaptively rank items for heterogeneous users, using contextual bandits and an upper confidence bound, achieving low regret and improving user satisfaction.
https://openreview.net/forum?id=e5admkWKgV
Compressor summary: Embodied AI (E-AI) is proposed as a key step toward Artificial General Intelligence (AGI), focusing on embodiment, cognitive architectures, and active inference to enhance AI's ability to communicate, collaborate, and coexist with humans.
https://openreview.net/forum?id=e3geukCBw6
Compressor summary: Momentor is a Video-LLM that can understand and locate specific video segments using a large-scale dataset called Moment-10M.
https://openreview.net/forum?id=e3Dpq3WdMv
Compressor summary: This study evaluates the trade-off between compression efficiency and trustworthiness in large language models using various techniques and dimensions, finding that quantization within a moderate bit range is more effective than pruning for achieving both goals.
https://openreview.net/forum?id=e1jPdRJeo7
Compressor summary: The paper introduces a new method called Zeroth-order Proximal Double Variance Reduction (ZPDVR) that reduces sampling and coordinate-wise variances in zeroth-order optimization, with less computational cost and better performance than existing methods.
https://openreview.net/forum?id=e0SKaKEEdr
Compressor summary: pcDEQ models improve DEQ models by ensuring positive weights and concave activation functions, providing theoretical guarantees for their fixed point and convergence, and performing well in language modeling and computer vision tasks.
https://openreview.net/forum?id=dztd61efGy
Compressor summary: SteerFair is a method to reduce biases in foundation models' QA capabilities by steering their internal representations away from spurious associations between input characteristics and correctness likelihood.
https://openreview.net/forum?id=dyfsPNuYCk
Compressor summary: The paper proposes a two-step purification method using diffusion models to remove noises in imperfect expert demonstrations for imitation learning, improving performance in real-world scenarios.
https://openreview.net/forum?id=dwWef5w2cR
Compressor summary: The paper proposes a Bayesian method called robust inverse graphics to infer 3D scenes from single images with unknown corruptions, using a strong scene prior and an uninformative uniform corruption prior, and shows that it outperforms other methods.
https://openreview.net/forum?id=duyl8sy8qV
Compressor summary: Slot Abstractors is a new approach that combines slot-based methods and relational abstraction to enable scalable abstract visual reasoning with many objects and relations.
https://openreview.net/forum?id=duRRoGeoQT
Compressor summary: Key points: - Transformers are widely used for sequence modeling but GSSMs have fixed-size latent state that does not depend on sequence length - The paper shows that GSSMs are less efficient and generalize worse than transformers on tasks that require copying from the input context - The paper evaluates pretrained language models and finds transformers significantly outperform GSSMs at copying and retrieving information from context Summary: The paper compares transformers and GSSMs, showing that transformers are superior for tasks that need copying from the input context.
https://openreview.net/forum?id=dtVlc9ybTm
Compressor summary: The paper proposes a new RL method to find high-reward samples in complex distributions by exploring the feasible manifold efficiently, with theory and experiments in images, biological sequences, and molecules.
https://openreview.net/forum?id=dslUyy1rN4
Compressor summary: The text argues that to advance robotics with sim-to-real reinforcement learning, researchers should focus on automating environment shaping procedures rather than tuning RL algorithms.
https://openreview.net/forum?id=dskLpg8WFb
Compressor summary: The paper proposes a graph contrastive learning method that preserves the graph community structure during augmentation, improving robustness and generalization.
https://openreview.net/forum?id=drjjxmi2Ha
Compressor summary: The paper proposes label smoothing to improve deep partial label learning classifiers and provides theoretical and empirical evidence for its effectiveness.
https://openreview.net/forum?id=dqpg8jdA2w
Compressor summary: Energy-based transition models capture complex real-world transitions and improve offline reinforcement learning performance.
https://openreview.net/forum?id=dqdctVbSfs
Compressor summary: The paper develops a new analysis of neural TD learning algorithms that improves the sample complexity and achieves an $ ilde{\mathcal{O}}(\epsilon^{-1})$ error bound under Markovian sampling.
https://openreview.net/forum?id=dplgaRn4Ae
Compressor summary: The paper proposes a new exploration by optimization approach with a hybrid regularizer to improve regret bounds in online decision-making problems under limited feedback, achieving nearly optimal performance in both stochastic and adversarial environments.
https://openreview.net/forum?id=dmfvHU1LNF
Compressor summary: The paper presents a new policy optimization algorithm (ACPO) for average-CMDPs with theoretical guarantees and experimental results showing its effectiveness in various challenging environments.
https://openreview.net/forum?id=dmHHVcHFdM
Compressor summary: Key points: - The paper proposes FABE, a new framework based on causal inference to protect language models from backdoor attacks. - FABE creates a 'front door' that maps out the actual causal relationships and filters out spurious associations. - FABE achieves state-of-the-art results in defending against various attack methods. Summary: FABE is a new framework that uses causal inference to defend language models from backdoor attacks by creating a 'front door' that separates legitimate and spurious associations, improving the defense effect significantly.
https://openreview.net/forum?id=dkdilv4XD4
Compressor summary: The balanced resonate-and-fire neuron (BRF) is an improved spiking neural network model that achieves higher performance, lower spike count, fewer parameters, faster convergence, and better stability than previous models.
https://openreview.net/forum?id=disVlUOH4b
Compressor summary: HOP is a novel multi-agent algorithm that efficiently adapts to co-players in mixed-motive environments by hierarchically modeling opponents' goals and using Monte Carlo Tree Search for planning.
https://openreview.net/forum?id=dhrNfAJAH6
Compressor summary: ELTA is a technique that enhances aesthetic image assessment by improving minority feature representation, aligning features and labels, and refining output distribution, especially for long-tailed datasets.
https://openreview.net/forum?id=dh8k41g775
Compressor summary: LQER is a method to reduce quantization errors in large language models, enabling near-lossless compression and improved downstream task performance with less hardware resources.
https://openreview.net/forum?id=dfR6FU53qk
Compressor summary: The text describes a method to improve long-term forecasting of dynamical systems using techniques from operator theory and statistics, with uniform error bounds on infinite time horizons.
https://openreview.net/forum?id=ddjRdm3wUW
Compressor summary: The paper analyzes implicit neural networks' spectral behavior using random matrix theory and proposes a method to design shallow explicit networks that match their kernel matrices.
https://openreview.net/forum?id=dcwUGaK9sQ
Compressor summary: The paper proposes DaGCN, a framework that improves Graph Convolutional Networks by handling nodes that do not fit the label smoothness assumption and are not well-represented by existing GCNs.
https://openreview.net/forum?id=dccRCYmL5x
Compressor summary: Equivariant Graph Neural Operator (EGNO) is a novel method that learns 3D dynamics as trajectories using equivariant temporal convolutions, outperforming existing methods in multiple domains.
https://openreview.net/forum?id=dbFEFHAD79
Compressor summary: The paper introduces a benchmark called MLLM-as-a-Judge to evaluate multimodal large language models, finding they perform well in pair comparisons but struggle with scoring evaluation and batch ranking tasks due to various biases and inconsistencies.
https://openreview.net/forum?id=da7MMwICjC
Compressor summary: RIS is a novel sampling method that reduces variance and improves convergence, performance, and uncertainty estimation in BNNs.
https://openreview.net/forum?id=dZsEOFUDew
Compressor summary: The text proposes a perspective to understand how pre-training language models with next-token prediction enables them to reason by aggregating indirect paths seen during pre-training on knowledge and reasoning graphs.
https://openreview.net/forum?id=dYDPcx78tm
Compressor summary: The paper analyzes how well offline data can be used for online decisions, finding near-optimal performance bounds based on function approximation and a new policy concept that covers all existing data-related notions.
https://openreview.net/forum?id=dWxb80a0TW
Compressor summary: The paper proposes a universal geometric graph representation for 3D molecular complexes and a Generalist Equivariant Transformer (GET) model to capture interactions between various molecule types using one model that preserves fine-grained information.
https://openreview.net/forum?id=dW29JZj0G5
Compressor summary: The paper proposes DARL, a simple decoder-only Transformer that learns strong visual representations for image generation by using tailored noise schedules and longer training in larger models.
https://openreview.net/forum?id=dVpFKfqF3R
Compressor summary: The paper proposes using categorical cross-entropy for training value functions in deep reinforcement learning, which improves performance and scalability across various domains.
https://openreview.net/forum?id=dVhrnjZJad
Compressor summary: The paper proposes factorized diffusion models for text-to-speech generation that disentangle speech into different attributes and generate them individually, leading to improved speech quality and naturalness.
https://openreview.net/forum?id=dV9QGostQk
Compressor summary: The paper introduces a flexible and scalable Bayesian framework for FANOVA models that can handle different sparsity levels and unify various methods, while enabling uncertainty quantification and novel model developments.
https://openreview.net/forum?id=dV9B9qFeGi
Compressor summary: The paper proposes a new loss function, M3G, that uses multi-marginal optimal transport theory to learn representations from multiple views, showing improved performance in self-supervised and multimodal tasks.
https://openreview.net/forum?id=dT6ZbSxh33
Compressor summary: The text discusses how generative AI impacts content creation, introduces a competition model to study the balance between humans and AI, and suggests a stable equilibrium is possible.
https://openreview.net/forum?id=dSrdnhLS2h
Compressor summary: The paper proposes a method called SubsCoRe that uses Gaussian process surrogate to control the cost and accuracy of SMO in VQE by adjusting the required accuracy and number of measurement shots.
https://openreview.net/forum?id=dQveBV9lZl
Compressor summary: The paper introduces Neural Walk-on-Spheres, a new neural network method for solving high-dimensional Poisson equations efficiently, with better accuracy, speed, and reduced memory usage compared to existing methods.
https://openreview.net/forum?id=dMhF96PfQi
Compressor summary: Semi-dual JKO is a scalable Wasserstein gradient flow model with reduced training complexity that achieves competitive results in image generation.
https://openreview.net/forum?id=dMOhgHNYAf
Compressor summary: BlobGEN is a text-to-image model that uses dense blob representations to capture fine-grained scene details, enabling better controllability and compositionality with large language models.
https://openreview.net/forum?id=dLojMSgSFW
Compressor summary: Independent evaluation of generative AI systems is crucial for safety, but current terms of service and research access programs discourage it; developers should provide legal and technical safe harbor for public interest research without fear of account suspension or legal reprisal.
https://openreview.net/forum?id=dJTChKgv3a
Compressor summary: The proposed in-context vectors (ICV) approach improves the efficiency and effectiveness of in-context learning for large language models, enabling them to follow demonstration examples better and handle diverse tasks more flexibly.
https://openreview.net/forum?id=dHXKCyaIkp
Compressor summary: The Deep Functional Factor Model (DF2M) is a Bayesian nonparametric model that combines the Indian Buffet Process, multi-task Gaussian Processes, and a deep kernel function to analyze high-dimensional functional time series, offering explainability and better predictive accuracy than conventional deep learning models.
https://openreview.net/forum?id=dGDFZM018a
Compressor summary: Key points: - Graph Neural Networks (GNNs) are powerful for homophilic graphs but not for heterophilic graphs with dissimilar node features - Signed Message Passing (SMP) is a widely used method to handle heterophilic graphs but has limitations - The paper proposes Multiset to Multiset GNN (M2M-GNN), a novel message-passing function that overcomes the limitations of SMP and performs better Summary: The paper introduces M2M-GNN, a new method for graph neural networks that handles heterophilic graphs with different node features better than existing methods like SMP.
https://openreview.net/forum?id=dFEeI51O5j
Compressor summary: MoNet is a modular network that learns task-specific decision-making processes without supervision, enabling effective and interpretable visual navigation in indoor environments.
https://openreview.net/forum?id=dBqHGZPGZI
Compressor summary: The study examines how direct preference optimization (DPO) reduces toxicity in pre-trained language models and reveals that it bypasses rather than removes capabilities learned from pre-training.
https://openreview.net/forum?id=dBMLtuKH01
Compressor summary: The authors propose scientific pragmatism, a balanced approach between scientific perfectionism and system-centric biases, to advance causal models and methods in knowledge generation and application.
https://openreview.net/forum?id=d5tJWH5yCi
Compressor summary: Key points: - The paper proposes a dynamic algorithm for maintaining decision trees under adversarial updates. - The algorithm guarantees good tree quality and fast update time. - The algorithm works for various metrics and types of decision trees, including boosted ones. Summary: The paper presents a fast and high-quality dynamic algorithm for updating different types of decision trees under data changes, with provable worst-case bounds.
https://openreview.net/forum?id=d5jXW2H4gg
Compressor summary: The paper proposes an approach to interpret higher-order interactions in complex ML models using the Shapley Interaction Index and shows its effectiveness with KernelSHAP-IQ.
https://openreview.net/forum?id=d5LURMSfTx
Compressor summary: InfiAgent-DABench is a benchmark to evaluate LLM-based agents on data analysis tasks using a format-prompting technique and an agent framework.
https://openreview.net/forum?id=d2vONO90Rw
Compressor summary: The authors propose a new method called supervised pinpoint tuning (SPT) that selectively fine-tunes specific modules in large language models to reduce sycophancy without compromising their general abilities.
https://openreview.net/forum?id=d2f2sCXQuI
Compressor summary: The paper proposes GRATH, a method to improve the truthfulness of large language models using out-of-domain question prompts and direct preference optimization.
https://openreview.net/forum?id=d2E2i5rJ4x
Compressor summary: The paper presents new algorithms for finding and decomposing dense subgraphs in graphs, improving on previous work in terms of iteration complexity, convergence rate, and practicality.
https://openreview.net/forum?id=d1P6GtRzuV
Compressor summary: The paper proposes a new type of temporal point process model, called Neural Jump-Diffusion Temporal Point Process (NJDTPP), that uses neural networks to parameterize its intensity dynamics and achieves better performance than existing models.
https://openreview.net/forum?id=cy3JBZKCw1
Compressor summary: The paper introduces a method called CA-TRIDE to improve image retrieval robustness against adversarial examples by addressing the limitations of existing deep metric learning approaches, and shows its effectiveness on three datasets.
https://openreview.net/forum?id=cxiqxDnrCx
Compressor summary: This paper introduces a new method, MIREL, for uncertainty estimation in weakly-supervised learning scenarios with multiple instances and weak annotations.
https://openreview.net/forum?id=cwIhvoTzuK
Compressor summary: The paper proposes an optimal algorithm for one-dimensional mean estimation under the add-remove model of differential privacy, showing that it performs similarly to the swap model and improves upon existing methods.
https://openreview.net/forum?id=coP4kPdhKr
Compressor summary: The paper proposes a fast dynamic spectral clustering algorithm for evolving graphs that can approximate cluster structures well under certain conditions.
https://openreview.net/forum?id=cmy38XZlJu
Compressor summary: The paper introduces a new distance for comparing probabilities that is faster, easier to estimate, and has more properties than the existing Wasserstein distance.
https://openreview.net/forum?id=cmD5E6ami4
Compressor summary: Symmetric replay training (SRT) improves sample efficiency in deep reinforcement learning for combinatorial optimization by using high-reward samples to explore under-explored symmetric regions without additional online interactions.
https://openreview.net/forum?id=cj5HbaX14p
Compressor summary: The paper proposes a new acquisition function called BOtied, which uses the CDF indicator to efficiently optimize multiple competing objectives using copulas.
https://openreview.net/forum?id=cit0hg4sEz
Compressor summary: FedKSeed is a method for federated full-parameter tuning of large language models that reduces communication cost by using zeroth-order optimization, random seeds, and probability-differentiated seed sampling, improving performance over existing methods.
https://openreview.net/forum?id=chhIZGqlUG
Compressor summary: The authors propose a new video format called Taylor video that extracts dominant motions from videos using Taylor expansion and show its effectiveness for action recognition with different architectures and modalities.
https://openreview.net/forum?id=chDpBp2P6b
Compressor summary: The paper proposes a new anomaly detection method using Non-Parametric Transformers to capture feature and sample dependencies and achieves state-of-the-art performance on 31 benchmark datasets.
https://openreview.net/forum?id=ccSSKTz9LX
Compressor summary: This paper studies how fine-tuning affects long-tail learning tasks and proposes LIFT, a lightweight fine-tuning algorithm that reduces training time, parameters, and improves performance.
https://openreview.net/forum?id=cc72Vnfvoc
Compressor summary: The authors present an optimization-based method to reconstruct datasets used for training random forests, using common libraries and showing that most methods are vulnerable to this attack.
https://openreview.net/forum?id=cbZTnjqIib
Compressor summary: This paper examines how constraining outputs of an ML model at inference time affects its generalization error and suggests choosing a proper loss function for this approach.
https://openreview.net/forum?id=cZTFxktg23
Compressor summary: The paper proposes a generative framework that models the topology of graphs by explicitly learning the final graph structures of the diffusion process, improving graph and molecule generation tasks.
https://openreview.net/forum?id=cZNuYKtoOZ
Compressor summary: The paper proposes a method to estimate continuous treatment effects using surrogate variables and labeled/unlabeled data, improving accuracy and addressing selection bias.
https://openreview.net/forum?id=cY9g0bwiZx
Compressor summary: The paper presents a new approach for multi-objective reinforcement learning that ensures fairness among multiple goals and improves performance over existing methods.
https://openreview.net/forum?id=cXBv07GKvk
Compressor summary: The paper argues that a new optimizer called IVON performs well for training large neural networks and has advantages over Adam in terms of predictive uncertainty, finetuning, and sensitivity to data.
https://openreview.net/forum?id=cXBPPfNUZJ
Compressor summary: The paper proposes a density estimation-based exploration method for 3-D environments using clustering on random or pre-trained representations, showing its effectiveness and integration of pre-trained biases in exploration.
https://openreview.net/forum?id=cVp8blEw2i
Compressor summary: The FESSNC is a fast neural controller for nonlinear systems that ensures stability and safety using heuristic learning, projection operators, and Hutchinson's trace estimator.
https://openreview.net/forum?id=cVkqItmYLQ
Compressor summary: This paper explores implicit neural representations using sampling theory, finding that $\mathrm{sinc}$ activations are optimal for shallow encodings and connecting them to dynamical systems.
https://openreview.net/forum?id=cUMOVfOIve
Compressor summary: The paper introduces SIN, a selective and interpretable method for normalizing time series data to improve forecasting accuracy.
https://openreview.net/forum?id=cU20finY8V
Compressor summary: Sharpness-aware minimization (SAM) improves generalization by discarding undesirable model biases through perturbed forgetting, which is related to information bottleneck principle and outperforms standard SAM on various benchmarks.
https://openreview.net/forum?id=cPsn9AcOYh
Compressor summary: The impact of QA fine-tuning data on factuality depends on the familiarity of facts, with lesser-known facts reducing factuality and better-known ones maintaining or improving it.
https://openreview.net/forum?id=cMige5MK1N
Compressor summary: Fed3R is a method for federated learning that works well with non-identical data distributions, is efficient, and can be fine-tuned with other algorithms.
https://openreview.net/forum?id=cHJAUdam3i
Compressor summary: The paper proposes MCD, a variational inference framework to discover causal models from time series data with different underlying causal structures, and shows its superior performance on synthetic and real datasets.
https://openreview.net/forum?id=cFDaYtZR4u
Compressor summary: CInA is a new self-supervised method that uses multiple unlabeled datasets for causal learning and enables zero-shot causal inference on unseen tasks with high accuracy, potentially paving the way for causal foundation models.
https://openreview.net/forum?id=cEJ9jNJuJP
Compressor summary: The paper questions the effectiveness of using machine learning to generate heatmaps for guiding Monte Carlo tree search in solving large-scale traveling salesman problems, and suggests future research directions.
https://openreview.net/forum?id=cBWVJh5Fvf
Compressor summary: AST-T5 is a novel pretraining method that uses Abstract Syntax Trees to improve code generation, transpilation, and understanding tasks, outperforming similar-sized language models in various scenarios.
https://openreview.net/forum?id=cAWbm9KRZO
Compressor summary: The LSC-transformation method improves language model alignment by emphasizing poorly-performing outputs, preventing underfitting and reward hacking, and allowing principled aggregation of multiple rewards.
https://openreview.net/forum?id=c9HddKGiYk
Compressor summary: The paper proposes a new Bayesian method to adjust network depth and width in dynamic architecture-based continual learning, achieving better or similar results than existing methods and working well for unsupervised learning too.
https://openreview.net/forum?id=c92KDfEZTg
Compressor summary: The paper presents AIM, a set of vision models inspired by Large Language Models that scale well with data and model size, and can be pre-trained on a 7 billion parameter model on 2 billion images for high ImageNet-1k performance.
https://openreview.net/forum?id=c8qWiNiqRY
Compressor summary: DiPmark is a watermarking technique that preserves the original content's distribution while being accessible and robust to token changes.
https://openreview.net/forum?id=c6rVlTKpb5
Compressor summary: The paper explores the challenges and solutions for hybrid reinforcement learning with observation-only offline data, and proposes an algorithm that performs well even without a reset model of the environment.
https://openreview.net/forum?id=c3ls5AVOw7
Compressor summary: The text discusses the importance of high-quality data for AI/ML models, the challenges of collecting such data, and how survey methodology can help improve data quality and reduce biases.
https://openreview.net/forum?id=c2CKmP9l5X
Compressor summary: The paper proposes DeepNeRAP, a method to learn a continuous neural field that encodes sound propagation dynamics in 3D spaces and infers room impulse response without direct ground truth access.
https://openreview.net/forum?id=c1AKcA6ry1
Compressor summary: The paper explores the theory and algorithms of aligning generative models with RLHF, proposes new methods that outperform existing ones, and shows their effectiveness on a large language model.
https://openreview.net/forum?id=c18noxRh3X
Compressor summary: A3S is a novel framework that improves active clustering performance by adjusting initial cluster results using Normalized mutual information gain and reducing human queries.
https://openreview.net/forum?id=c0LoolDFw4
Compressor summary: The paper explores how to control language models with different concepts beyond truthfulness and evaluates their effectiveness using a new metric and extensive experiments.
https://openreview.net/forum?id=bzNwexOPWm
Compressor summary: Key points: - Language models make errors and suffer from catastrophic forgetting when updated with corrected instances - The goal is to predict which upstream examples will be forgotten for better replay control and interpretability - A partially interpretable model based on logit scores performs well on BART but not on T5 - A black-box classifier based on inner products of representations outperforms the interpretable model - Replaying forecasted forgotten examples reduces forgetting and shows practical utility Summary: The paper proposes models to predict which upstream examples will be forgotten by language models after update, using either interpretable or black-box methods, and shows that replaying these examples improves performance and reduces forgetting.
https://openreview.net/forum?id=byxXa99PtF
Compressor summary: Input clarification ensembling is a framework for decomposing the uncertainty of large language models into aleatoric (data) and epistemic (model) components, improving reliability and interpretability.
https://openreview.net/forum?id=bylZbZOsGA
Compressor summary: The paper presents a neuro-symbolic approach to automatically translate informal Euclidean geometry proofs into formal theorems using theorem provers, large language models, and semantic evaluation.
https://openreview.net/forum?id=byAXJTk0LH
Compressor summary: SNORE is a new PnP algorithm that applies the denoiser only on images with appropriate noise levels, improving image restoration results.
https://openreview.net/forum?id=bwZlD7mYoa
Compressor summary: The paper proposes a new method to detect periodic patterns in noisy time series data without labels or augmentations and shows significant improvements over existing methods.
https://openreview.net/forum?id=bvPYroQgc3
Compressor summary: The study examines how optimal ridge regularization and risk behave in out-of-distribution prediction scenarios, showing that negative regularization can be optimal and the tuned risk depends on data aspect ratio.
https://openreview.net/forum?id=buW1Bi6XFw
Compressor summary: Randomly shuffling feature vectors among nodes of the same class improves graph neural network performance by reducing the dependence between graph topology and features.
https://openreview.net/forum?id=btYeH65fI3
Compressor summary: The paper studies how adversarial attacks affect linear regression models and finds the best balance between robustness and accuracy for different scenarios.
https://openreview.net/forum?id=bqgtkBDkNs
Compressor summary: The paper discusses challenges in evaluating Bayesian Causal Discovery methods due to uncertainty in inferred causal graphs and proposes factors to consider for better assessment.
https://openreview.net/forum?id=bq1JEgioLr
Compressor summary: SciBench is a benchmark suite for testing large language models on complex scientific problems across math, chemistry, and physics domains, revealing their current limitations and areas for improvement.
https://openreview.net/forum?id=bplNmU2ROC
Compressor summary: BaM is a new black-box variational inference method that uses a score-based divergence and optimizes Gaussian variational families with full covariance matrices, which converges faster and better than ELBO-based methods.
https://openreview.net/forum?id=bmeUeCUMHA
Compressor summary: A possible summary is: The authors propose a novel sequential goodness-of-fit testing method that adapts to data complexity and maintains statistical power by using a betting strategy.
https://openreview.net/forum?id=blzDxD6bKt
Compressor summary: The paper proposes a new nonlinear filtering method based on Brenier optimal transport that uses neural networks and stochastic optimization to handle degenerate likelihoods, high-dimensional states, and multi-modal distributions.
https://openreview.net/forum?id=blGpu9aGs6
Compressor summary: Key points: - New algorithm (AR) for accurate tree-based models with recourse actions - Ensures existence of actions by constraining tree growth - Uses adversarial training and greedy algorithm - Applies to random forest and improves accuracy and efficiency Summary: The paper introduces AR, a new algorithm that learns accurate tree models with guaranteed recourse actions using adversarial training and a greedy approach. It works on random forests and outperforms baselines in accuracy and efficiency.
https://openreview.net/forum?id=biE1uHyG0l
Compressor summary: This paper studies how two agents can help a central server estimate a high-dimensional covariance matrix by communicating limited information about disjoint samples.
https://openreview.net/forum?id=bgP8Rxv2eB
Compressor summary: Key points: - CMLL problem: multiple true labels, unreliable labels from annotators - Existing methods: focus on inferring true labels, not predicting - New method: unbiased risk estimator based on transition matrices, decoupled autoencoder to exploit label correlations - Generalization error bound for convergence Summary: The paper proposes a new method for CMLL that uses an unbiased risk estimator and a decoupled autoencoder to handle unreliable labels and improve prediction.
https://openreview.net/forum?id=bfQCO9Vqhk
Compressor summary: The paper proposes sparsity-constrained optimal transport methods for learning sparse transport plans with efficient algorithms and theory.
https://openreview.net/forum?id=beXQVQorse
Compressor summary: Teacher-Student Bayesian Optimization (TSBO) is a novel semi-supervised learning approach that uses unlabeled data, teacher and student models to minimize labeled data queries and improve sample-efficiency in global optimization tasks.
https://openreview.net/forum?id=bdKaQmrM81
Compressor summary: The paper presents coordinate descent algorithms for various matrix manifolds that update fewer variables per iteration than Riemannian optimization while maintaining feasibility and providing low cost and efficiency.
https://openreview.net/forum?id=bcN7KSB2YS
Compressor summary: The paper studies games with state constraints and one-sided information, and provides theoretical results for computing behavioral strategies and belief manipulation in such games.
https://openreview.net/forum?id=bZNH0SU37Y
Compressor summary: The effect of temporal aggregation on non-temporal causal discovery depends on the consistency criterion used, the degree of nonlinearity, and the presence of partial linearity or prior information.
https://openreview.net/forum?id=bZ4fzw1iz7
Compressor summary: SnD is a private inference framework that splits LLMs and adds noise to embeddings before transmitting them to servers, enhancing privacy while maintaining performance.
https://openreview.net/forum?id=bYRYb7DMNo
Compressor summary: The paper introduces the Time Series Transformer (Timer), a large language model pre-trained on a curated dataset of heterogeneous time series, which can perform various tasks such as forecasting, imputation, and anomaly detection.
https://openreview.net/forum?id=bX3J7ho18S
Compressor summary: The paper proposes a method to estimate LLM-generated text in scientific peer reviews and finds that 6.5%-16.9% of the review text could be modified by LLMs, with higher occurrence in low confidence reviews and near deadline submissions.
https://openreview.net/forum?id=bWZKvF0g7G
Compressor summary: The text discusses a new vision-language safe instruction-following dataset called VLGuard that helps improve the safety of large language models without compromising their helpfulness.
https://openreview.net/forum?id=bWUU0LwwMp
Compressor summary: The paper introduces TrustLLM, a study on trustworthiness in large language models, evaluating 16 mainstream models across eight dimensions and finding that proprietary models generally outperform open-source ones, but some open-source models come close to them.
https://openreview.net/forum?id=bWNPx6t0sF
Compressor summary: The paper analyzes various methods for fine-tuning language models with preference data and finds that approaches using on-policy sampling and negative gradient outperform offline and maximum likelihood objectives, which are unified under mode-seeking objectives for categorical distributions.
https://openreview.net/forum?id=bVIcZb7Qa0
Compressor summary: The paper proposes controlled decoding (CD), a modular solver that uses a prefix scorer to learn a value function and control the generation of a frozen base model, achieving effective alignment of language models with multiple rewards.
https://openreview.net/forum?id=bV9yT24t9B
Compressor summary: The paper introduces a new method for generating code called self-infilling, which can create context and content simultaneously, improving output quality and regularizing the generation process.
https://openreview.net/forum?id=bULHOW1RXM
Compressor summary: DMS is a method that efficiently searches for optimal network width and depth, improving performance on various tasks such as image classification, object detection, and language modeling.
https://openreview.net/forum?id=bPsohGR6gD
Compressor summary: The paper introduces a new model for bandits where arms' rewards depend on each other through a graph, studies its optimal and suboptimal policies, and presents regret minimization algorithms with structured graphs and no-regret properties.
https://openreview.net/forum?id=bOhzU7NpTB
Compressor summary: Modular-DCM is an efficient algorithm that uses pre-trained models to sample from identifiable causal queries in high-dimensional data, outperforming baselines and handling latent confounders.
https://openreview.net/forum?id=bNgAdyv7ZP
Compressor summary: The paper proposes a new measure called the coefficient of prescriptiveness to compare optimization techniques that use side information for better decisions and introduces a model with this measure, which is solved by a bisection algorithm.
https://openreview.net/forum?id=bM2s12t4hR
Compressor summary: This paper proves that strong watermarking schemes for generative models are impossible under realistic assumptions and shows an efficient attack on three existing schemes using a "quality" and "perturbation" oracle.
https://openreview.net/forum?id=bJbSbJskOS
Compressor summary: Genie is a large-scale generative model that can create diverse virtual environments from text or images and can be controlled by user actions without any supervision.
https://openreview.net/forum?id=bID9PiBFpT
Compressor summary: The paper proposes a new algorithm for reinforcement learning that considers risk as measured by asymptotic variance and shows it converges in finite time.
https://openreview.net/forum?id=bBzlapzeR1
Compressor summary: The paper examines how importance re-weighting improves kernel ridge regression in high dimensions with covariate shifts by analyzing the bias-variance trade-off and providing asymptotic expansions of kernels under covariate shift.
https://openreview.net/forum?id=bBkQ51PmjC
Compressor summary: The paper proposes a novel learning-based approach for solving Quadratic Assignment Problems (QAPs) using a Solution AWare Transformer (SAWT) architecture that encodes facility and location nodes separately, enabling scalability to larger problem sizes.
https://openreview.net/forum?id=b9uHveqszc
Compressor summary: The paper analyzes how the $D^\alpha$ seeding algorithm performs better than standard $k$-means for any $\alpha>2$, and gives theoretical and empirical evidence for this improvement.
https://openreview.net/forum?id=b9VfvegTEO
Compressor summary: FALCON is an unsupervised method that uses coarse labels to discover and relate fine-grained classes in image and single-cell classification tasks, improving performance significantly.
https://openreview.net/forum?id=b89JtZj9gm
Compressor summary: Modern Text-to-Image generators like Stable Diffusion can improve the robustness and generalization of neural image classifiers by simulating interventions over environmental factors in the training data.
https://openreview.net/forum?id=b6yHkQpSwZ
Compressor summary: The paper proposes a novel way to convert graphs into sets and use set encoders like Transformers to learn from them, improving the expressivity and performance of Graph Neural Networks.
https://openreview.net/forum?id=b6rA0kAHT1
Compressor summary: Key points: - The paper proposes an algorithmic framework for multi-turn RL with LLMs that balances flexibility and efficiency. - The framework, ArCHer, combines a high-level value function learner and a low-level token-by-token policy learner. - ArCHer outperforms prior methods in sample efficiency and performance on multi-turn LLM tasks. Summary: The paper introduces ArCHer, an algorithmic framework for multi-turn RL with LLMs that uses a high-level value function and a low-level token-by-token policy to optimize long-term objectives efficiently and effectively.
https://openreview.net/forum?id=b6AwZauZPV
Compressor summary: The paper proposes a probabilistic subgoal representation function for hierarchical reinforcement learning using Gaussian Processes, which improves performance in various tasks and environments with stochastic uncertainties and diverse rewards.
https://openreview.net/forum?id=b3pYoZfcoo
Compressor summary: SAPS is a novel algorithm for conformal prediction that reduces prediction set size by discarding probability values except for the maximum softmax probability, while preserving uncertainty information and improving conditional coverage rates.
https://openreview.net/forum?id=b2D9PBNNQ2
Compressor summary: The paper proposes a simple algorithm to represent heavy matrix entries with low bit-width integers and achieve efficiency gains in GEMM operations for Transformer-based models.
https://openreview.net/forum?id=b1iurBHDck
Compressor summary: The authors propose a novel algorithm that uses a multimodal variational autoencoder to reconstruct nonlinear dynamical systems from various types of data, including symbolic data, in a generative framework.
https://openreview.net/forum?id=b1YQ5WKY3w
Compressor summary: The paper investigates whether in-context learning in large language models is equivalent to Bayesian inference using the martingale property, and finds evidence against this hypothesis.
https://openreview.net/forum?id=b0lxGL2n3d
Compressor summary: The paper presents a new machine learning framework that can generate high-quality masks for chip design without using inverse lithography solvers, achieving better performance than existing methods.
https://openreview.net/forum?id=axwrD8F1yq
Compressor summary: The text proposes a novel method for active domain adaptation that focuses on improving individual categories without harming others by identifying the most important unlabeled data samples using influence function analysis.
https://openreview.net/forum?id=axl3FAkpik
Compressor summary: The Binoculars method accurately detects machine-generated text from various large language models using simple calculations with two pre-trained models, achieving state-of-the-art performance without any training data or model-specific modifications.
https://openreview.net/forum?id=awo5H10K6v
Compressor summary: The algorithm uses LLMs to segment trajectories and then merges them using variational inference and an auxiliary objective to discover reusable skills for agents in different environments.
https://openreview.net/forum?id=aw6L8sB2Ts
Compressor summary: The paper studies how mixing real and synthetic data in generative models affects data distributions, derives bounds on the TV distance between them, and reveals a phase transition point where the distance declines.
https://openreview.net/forum?id=asJTE8EBjg
Compressor summary: The study reveals how large language models represent and decode beliefs, showing their importance for social reasoning in various tasks.
https://openreview.net/forum?id=arwP5FA2dO
Compressor summary: Key research challenges in fusion energy that could benefit from Machine Learning applications are discussed, highlighting six areas where ML can contribute to advancing fusion as a carbon-free energy source.
https://openreview.net/forum?id=ar174skI9u
Compressor summary: The paper proposes a novel attention-based encoder for solving the min-max vehicle routing problem, which improves the efficiency and optimality of sequential planning by decoupling customer partition and navigation tasks and using an agent-permutation-symmetric loss function.
https://openreview.net/forum?id=apxON2uH4N
Compressor summary: The paper proposes an efficient algorithm for privacy-preserving word embedding using homomorphic encryption and coded inputs.
https://openreview.net/forum?id=aoAPOOtN9E
Compressor summary: The paper introduces Thought Rollback, a new reasoning framework for LLMs that allows them to adaptively build thought structures and revise mistaken ones to solve problems better under hallucinations.
https://openreview.net/forum?id=anM1M5aoM8
Compressor summary: EvoluNet is a novel framework for non-IID transfer learning on dynamic graphs that leverages temporal encoding and domain unification modules to improve generalization performance.
https://openreview.net/forum?id=amRSBdZlw9
Compressor summary: The authors propose developing large tabular models (LTMs) that can contextualize multiple datasets, which could have significant impacts on various fields and tasks involving tabular data.
https://openreview.net/forum?id=akyElNlUVA
Compressor summary: FedLMT and pFedLMT are frameworks for federated learning that address resource heterogeneity among clients using pre-factorized low-rank models, improving model accuracy and reducing costs.
https://openreview.net/forum?id=aksdU1KOpT
Compressor summary: The paper proposes Multi-layer Rehearsal Feature Augmentation (MRFA) to improve generalization and reduce catastrophic forgetting in Class-Incremental Learning by optimizing the all-layer margin on rehearsal samples.
https://openreview.net/forum?id=aiz79FxjaI
Compressor summary: The paper introduces dependent Bayes optimality and a deferral principle for learning to defer frameworks that exploit the dependence between models and experts, and propose a novel consistent surrogate loss based on these concepts.
https://openreview.net/forum?id=ahEm3l2P6w
Compressor summary: The paper introduces OWL, a novel pruning method for large language models that uses non-uniform layerwise sparsity ratios based on outlier features to improve performance and reduce model size.
https://openreview.net/forum?id=ah1BlQcLv4
Compressor summary: Transformers can learn gradient descent and non-linear functions by implementing them in function space, enabling efficient learning for various non-linear tasks and activation choices.
https://openreview.net/forum?id=afnyJfQddk
Compressor summary: The paper introduces Gaussian processes on cellular complexes, a generalization of graphs that captures polyadic relations, and proposes two new kernels to model interactions between vertices, edges, and higher-order cells.
https://openreview.net/forum?id=aeXRBnLoPP
Compressor summary: The paper introduces Accelerated Policy Gradient (APG), a method to improve convergence rates in reinforcement learning by adapting Nesterov's accelerated gradient to policy optimization, and proves its theoretical convergence properties.
https://openreview.net/forum?id=ad5I6No9G1
Compressor summary: The paper explains why models can improve on modular addition problem even after overfitting, and suggests that this phenomenon is due to a transition from kernel-like behavior to limiting behavior of gradient descent on deep networks.
https://openreview.net/forum?id=acTLXagzqd
Compressor summary: This paper introduces a new autoencoder framework that preserves the data's geometric structure using a similarity graph and Riemannian geometry, leading to better latent representation learning.
https://openreview.net/forum?id=aaeJpJw5Ur
Compressor summary: The text proposes Multi-Agent Socialized Collaboration (MASC), a method for achieving socialized learning in multi-agent systems, which prioritizes original expert classes' accuracy while acquiring new abilities.
https://openreview.net/forum?id=aZnZOqUOHq
Compressor summary: The paper proposes a deep learning method for finding Lagrangian Multipliers that improves bounds on Mixed Integer Linear Programs by using graph neural networks to encode and decode relaxed constraints.
https://openreview.net/forum?id=aXD94eATtT
Compressor summary: Adaptive decoding is a mechanism that helps language models choose better candidates for tokens during generation by increasing confidence based on an entropy-based metric.
https://openreview.net/forum?id=aVqqoFAavs
Compressor summary: The paper proposes a method to extend neural operators for learning solutions of PDEs to non-equispaced point distributions by efficiently evaluating spectral transformations, achieving faster training and comparable or improved accuracy.
https://openreview.net/forum?id=aRZjRj41WQ
Compressor summary: The text discusses how self-attention mechanisms in machine learning can be improved by reducing localization and addressing rank and entropy collapses for better model performance.
https://openreview.net/forum?id=aR3uxWlZhX
Compressor summary: The paper proposes UP2ME, a general-purpose framework for multivariate time series tasks that combines univariate pre-training and multivariate fine-tuning to achieve state-of-the-art performance in forecasting, imputation and anomaly detection.
https://openreview.net/forum?id=aQl4xiwVBc
Compressor summary: The paper proposes PRepBN and SLA modules to improve efficiency of transformers for computer vision and natural language tasks, achieving lower latency and similar performance compared to existing methods.
https://openreview.net/forum?id=aPhwhueqjR
Compressor summary: The paper presents a method to adapt any convex optimization algorithm to work with inexact first-order information, without knowing the algorithm's details, and applies to various types of algorithms and nonconvex problems.
https://openreview.net/forum?id=aPVwOAr1aW
Compressor summary: The paper proposes a multi-embedding design to address embedding collapse, which restricts scalability in large recommendation models, and shows its effectiveness through experiments.
https://openreview.net/forum?id=aP0H8A1ywk
Compressor summary: The paper studies the Lipschitz properties of self-attention and masked self-attention in Transformers, showing how sequence length and layer normalization affect their bounds, and proposes a novel mean-field framework for masked self-attention.
https://openreview.net/forum?id=aLSA3JH08h
Compressor summary: The text discusses improving offline optimization by developing a sensitivity measurement for surrogate models and using it to create a regularizer that enhances optimization performance.
https://openreview.net/forum?id=aK1FyEP2Sn
Compressor summary: CokeNet is a novel spectral density function for spectral kernel networks that enhances diversity and captures complex data dependencies, leading to better machine learning performance.
https://openreview.net/forum?id=aGBpiEcB8z
Compressor summary: BayOTIDE is an online imputation method that models multivariate time series as a combination of low-rank temporal factors with different patterns, using Gaussian Processes and state-space priors.
https://openreview.net/forum?id=aECamk9izk
Compressor summary: Key points: - BNNs with high-dimensional parameters have multi-modal posterior distributions - SGMCMC with cyclical learning rate scheduling can explore them but is computationally expensive - The paper proposes a meta-learning strategy to build SGMCMC that efficiently explores the posteriors - The algorithm transfers well to various tasks and improves sampling efficiency Summary: The paper introduces a meta-learning method for building SGMCMC that can efficiently explore multi-modal posterior distributions of BNNs, achieving better performance on image classification and other tasks.
https://openreview.net/forum?id=aC1LSa4nXs
Compressor summary: ConfDiff is a force-guided diffusion model for protein conformation generation that incorporates physical prior knowledge to produce diverse and accurate conformations, outperforming existing methods.
https://openreview.net/forum?id=aA2326y3hf
Compressor summary: This paper studies how looped transformer networks with extra attention heads can simulate various graph algorithms and proves their theoretical properties.
https://openreview.net/forum?id=a9bzTv9SzO
Compressor summary: Rolling Diffusion is a novel method for denoising temporal data that adapts to uncertainty by adding more noise to later frames, outperforming standard diffusion methods in complex scenarios like video prediction and fluid dynamics.
https://openreview.net/forum?id=a8ZpjLJuKk
Compressor summary: The authors develop theory to understand critical windows, narrow time intervals in image generation when specific features emerge, and use the framework to analyze and diagnose issues in diffusion models.
https://openreview.net/forum?id=a8QpoEJCRI
Compressor summary: SurfPro is a method that generates functional proteins with desired surfaces and biochemical properties by encoding the geometric shape and biochemical features of a protein surface and decoding an amino acid sequence.
https://openreview.net/forum?id=a7MW5kFFOf
Compressor summary: The paper presents a new method for selecting important features and classifying functional data in scenarios with categorical responses and multivariate longitudinal features, which is faster and more accurate than existing methods.
https://openreview.net/forum?id=a6wCNfIj8E
Compressor summary: The paper proposes functional reward encoding (FRE), a method to pre-train a generalist agent from unlabeled data using a transformer-based variational auto-encoder, enabling it to solve new tasks in a zero-shot manner with minimal supervision.
https://openreview.net/forum?id=a6366lEzbX
Compressor summary: The text discusses the difficulty of quantifying realism in generated data and proposes a novel concept called universal critic that could help solve this problem without adversarial training.
https://openreview.net/forum?id=a3XFF0PGLU
Compressor summary: The paper presents a dual-agent reward shaping method that improves reinforcement learning's sample efficiency and stability in sparse-reward tasks by generating auxiliary reward signals.
https://openreview.net/forum?id=a2uFstsHPb
Compressor summary: The authors propose a novel multi-task learning method that uses low-rank matrices and orthogonal regularization to efficiently learn the Pareto front and achieve better performance, especially on large datasets.
https://openreview.net/forum?id=a1Olc2QhPv
Compressor summary: The paper develops a PAC-Bayesian error bound for linear time-invariant stochastic state-space models, which are used in control engineering and econometrics, and have applications in recurrent neural networks.
https://openreview.net/forum?id=a1GvTbadqA
Compressor summary: The paper introduces a new Thompson sampling variant called Varational Inference TS (VITS) that uses Gaussian Variational Inference for efficient posterior approximation and achieves sub-linear regret in linear contextual bandits.
https://openreview.net/forum?id=a0XiA6v256
Compressor summary: The authors show that diffusion models can estimate the intrinsic dimension of data manifolds by approximating their normal bundles using the score function, which points towards the manifold when noise is low.
https://openreview.net/forum?id=ZzXNCQGzqT
Compressor summary: Ditto is a framework that enables efficient and secure quantization-aware inference for Transformers using multi-party computation techniques, reducing computation and communication overhead.
https://openreview.net/forum?id=ZzFTrzo0Cp
Compressor summary: MC-GTA is a novel clustering algorithm that uses feature similarity and metric autocorrelation to improve performance, stability, and computational efficiency for various data analysis tasks.
https://openreview.net/forum?id=ZzCY0fRver
Compressor summary: The paper presents a framework that uses satellite data to estimate canopy heights globally with improved accuracy compared to existing methods, enabling better ecological analyses.
https://openreview.net/forum?id=ZxDqSBgFSM
Compressor summary: The paper proposes a method called Federated Graph Rationalization (FedGR) that uses anti-shortcut augmentations to generate explanations for predictions made by graph neural networks in federated learning settings.
https://openreview.net/forum?id=ZwrfsrCduj
Compressor summary: The paper proposes a method to distinguish between task-irrelevant visual distractors using an Implicit Action Generator (IAG) that learns the behavior of distractors and improves performance on various visual control tasks.
https://openreview.net/forum?id=ZwUThOE7Zc
Compressor summary: Machine learning in autonomous weapons systems increases the risk of low-intensity conflicts, arms race, and undermines geopolitical stability and AI research transparency.
https://openreview.net/forum?id=Zw7TcnTmHj
Compressor summary: The paper investigates transfer learning in high-dimensional linear regression with benign overfitting, showing how overparameterized models behave under different types of covariate shifts.
https://openreview.net/forum?id=Zw52bJCZXc
Compressor summary: The paper introduces two new stochastic Frank-Wolfe methods for optimization problems with structured constraints that have better convergence guarantees and avoid issues of large batches or full gradients.
https://openreview.net/forum?id=ZvJ2lQQKjz
Compressor summary: MERL is a multimodal learning framework for ECGs that uses reports and text prompts to classify heart diseases without training data, outperforming eSSL methods.
https://openreview.net/forum?id=ZvFLbEPv6x
Compressor summary: The study proposes a stealthy backdoor attack, SilentBadDiffusion, that can induce copyright infringement in text-to-image diffusion models without controlling their training, and shows its effectiveness on various model architectures.
https://openreview.net/forum?id=ZtOXZCTgBa
Compressor summary: SeMOPO is a new model-based offline RL approach that separates latent states into endogenous and exogenous parts, reducing bias in uncertainty estimation when dealing with complex distractors.
https://openreview.net/forum?id=ZtMqsSkIHX
Compressor summary: The text discusses modifying a game's rewards to achieve a specific policy goal, studying the conditions for success and proposing an algorithm to solve this problem efficiently.
https://openreview.net/forum?id=Zsz9Pdfvtg
Compressor summary: Key points: - Cross-modal transformers integrate different modalities for vision tasks - GeminiFusion is a pixel-wise fusion approach that combines intra-modal and inter-modal attentions - GeminiFusion adapts to the complexity of the input and achieves superior performance on various multimodal tasks Summary: GeminiFusion is a novel cross-modal transformer that fuses different modalities using aligned representations and adaptive attentions, achieving high performance on image-to-image translation, 3D object detection, and semantic segmentation.
https://openreview.net/forum?id=Zs3qW8Njov
Compressor summary: Smoothing Proximal Gradient Methods (SPGM) are explored as solutions to nonsmooth sparsity constrained optimization problems, with two variants (SPGM-IHT and SPGM-BCD) showing improved performance over existing methods in theory and practice.
https://openreview.net/forum?id=ZrM67ZZ5vj
Compressor summary: The paper proposes a novel method for building language-conditioned agents using vision-language models by first finding an optimal environment configuration and then using a goal-conditioned policy to reach it, achieving better zero-shot generalization than multi-task RL baselines.
https://openreview.net/forum?id=Zos5wsaB5r
Compressor summary: The paper proposes a method for online Vision-and-Language Navigation that adapts model parameters using fast-slow tests and shows improved performance on four benchmarks.
https://openreview.net/forum?id=ZoTIdyExx6
Compressor summary: The text studies an adaptive online optimization algorithm for online learning in changing environments and shows how it improves regularization and uncertainty prediction.
https://openreview.net/forum?id=Zo9zXdVhW2
Compressor summary: AWaVO is a novel method for reinforcement learning that provides interpretability through convergence guarantee, training transparency, and intrinsic decision-interpretation, achieving good performance in simulation and real-world quadrotor tasks.
https://openreview.net/forum?id=Zn44XGFGam
Compressor summary: The paper analyzes the performance of two-layer ReLU networks with weight decay regularization and their convex relaxations, showing that random training data can lead to a small optimality gap and improved convergence rates for local gradient methods.
https://openreview.net/forum?id=ZeF75iQcAc
Compressor summary: The paper presents new acceleration mechanisms for minimax optimization and fixed-point problems with the same optimality as existing anchor-based methods, but different characteristics.
https://openreview.net/forum?id=ZdqiT0McON
Compressor summary: The paper derives generalization bounds for the clean-label backdoor attack scenario and proposes a new attack method using a combination of adversarial noise and indiscriminate poison.
https://openreview.net/forum?id=ZdSe1qnuia
Compressor summary: The paper proposes new score-based methods for identifying causal structures involving latent variables, addressing challenges of existing constraint-based methods and providing identifiability guarantees.
https://openreview.net/forum?id=ZctlF8RlV4
Compressor summary: The paper introduces a novel optimization framework for one-shot structured pruning that improves efficiency and accuracy on vision and language models, including very large ones with tens of billions of parameters.
https://openreview.net/forum?id=Zc22RDtsvP
Compressor summary: The paper introduces MagicLens, a self-supervised image retrieval model that supports open-ended text instructions to find images with rich relations beyond visual similarity by using implicit relations from web pages.
https://openreview.net/forum?id=ZZ7UKgK4c1
Compressor summary: The paper introduces a new dataset and prompting approach to study how AI can understand characters' mental states in movies, and shows that existing models perform worse than humans in this task.
https://openreview.net/forum?id=ZXsNkm3bxu
Compressor summary: The text proposes a framework for generating synthetic data from distributed sources using secure multi-party computation and differential privacy, allowing data sharing without entrusting raw data to a central entity.
https://openreview.net/forum?id=ZVmMV3AHjC
Compressor summary: The paper proposes a model-free algorithm for two-player Markov games that achieves optimal sample complexity, using variance reduction and reference value functions.
https://openreview.net/forum?id=ZUXvpIrz5l
Compressor summary: SAFE-T is an adaptive dose-finding procedure that uses Bayesian optimization to learn non-parametric models for toxicity and efficacy, while satisfying safety constraints and improving utility for heterogeneous participants in early phase clinical trials.
https://openreview.net/forum?id=ZTN866OsGx
Compressor summary: MorphGrower is a learning-based method that mimics neuron growth to generate realistic and valid neural morphologies, outperforming previous approaches.
https://openreview.net/forum?id=ZSQAf5YlvN
Compressor summary: The optimal batch size for training two-layer neural networks with SGD depends on the target's hardness and information exponent, and a new protocol called Correlation loss SGD can improve training speed beyond traditional batch sizes.
https://openreview.net/forum?id=ZRMQX6aTUS
Compressor summary: The paper studies SGD algorithms with arbitrary data ordering and improves convergence guarantees for incremental gradient and single shuffle SGD.
https://openreview.net/forum?id=ZQcqXCuoxD
Compressor summary: The paper introduces a flexible and dynamic message-passing framework for graph neural networks, where nodes can choose to listen, broadcast, or isolate, enabling more effective exploration of the graph topology.
https://openreview.net/forum?id=ZPiEIhQpos
Compressor summary: The paper provides a convergence guarantee for Consistency Models, one-step generative models that can match Diffusion Models in sample quality.
https://openreview.net/forum?id=ZMgpE58PMj
Compressor summary: The authors propose a new method to find the best position and orientation for adsorbates on a slab, which can accelerate the discovery of novel catalysts by using denoising diffusion, machine learning force field, and DFT.
https://openreview.net/forum?id=ZGEICuuUJo
Compressor summary: The text formalizes the least-privilege principle for machine learning and proves its trade-off between utility and leakage, showing it is not possible to learn highly useful representations that prevent inference of unrelated attributes.
https://openreview.net/forum?id=ZFYBnLljtT
Compressor summary: This paper investigates how different tokenizer designs affect the performance of language models for code generation tasks, and provides recommendations for optimizing them.
https://openreview.net/forum?id=ZFRrOiZruJ
Compressor summary: The paper presents a new ATS framework that learns from data to create optimal personalized exams, reducing errors and questions while maintaining accuracy.
https://openreview.net/forum?id=ZAW37OZ6ig
Compressor summary: The paper introduces the pix2emb method for object location modeling in LMMs, enabling them to handle various multimodal tasks using different location formats.
https://openreview.net/forum?id=Z7MzVDFWDV
Compressor summary: The paper proposes a method to identify and estimate a parameter in missing not at random data by fusing it with auxiliary data that is missing at random.
https://openreview.net/forum?id=Z2LH6Va7L2
Compressor summary: UniFilter is a novel GNN that adapts polynomial bases to the heterophily degree of a graph, improving convolution and propagation, and facilitating graph explanation.
https://openreview.net/forum?id=Z19JQ6WFtJ
Compressor summary: The paper proposes a method that uses Large Language Models (LLM) and self-alignment to learn reward functions more efficiently for robot skills without human guidance.
https://openreview.net/forum?id=Z0S6fUdW68
Compressor summary: The paper proposes two algorithms for adversarial corruption in model-based reinforcement learning, one for online and one for offline settings, and provides theoretical guarantees for their performance.
https://openreview.net/forum?id=YxmcEfcgp3
Compressor summary: The paper studies how imposing the axis-aligned constraint on soft trees affects their learning behavior and explores various tree architectures using Neural Tangent Kernel analysis and Multiple Kernel Learning.
https://openreview.net/forum?id=Ywl6pODXjB
Compressor summary: Transolver uses Physics-Attention to learn intrinsic physical states from discretized geometries and improve solving partial differential equations.
https://openreview.net/forum?id=YvPNwLedpQ
Compressor summary: The paper proposes an algorithm for online convex optimization with arbitrary delays in non-stationary environments and shows its performance bounds and improvements over existing methods.
https://openreview.net/forum?id=YvAyOYeGlo
Compressor summary: Possible summary: The paper introduces AND, a framework that explains acoustic neurons' behaviors and features using natural language summaries from large language models.
https://openreview.net/forum?id=Yug1IEkvcb
Compressor summary: The paper proposes two model-free algorithms for learning optimal robust policies in high-dimensional systems using $\phi$-divergences, one that uses only historical data and another that combines historical and online sampling.
https://openreview.net/forum?id=YuNFJSEkTi
Compressor summary: CasCast is a cascaded framework that uses a diffusion transformer to improve precipitation nowcasting and forecasting of extreme events.
https://openreview.net/forum?id=YuGnRORkJm
Compressor summary: The paper proposes a Sample Average Approximation method for Conditional Stochastic Optimization with dependent data and provides theoretical guarantees of its consistency, sample complexity, and finite sample guarantee under mild conditions.
https://openreview.net/forum?id=Yqj3DzIC79
Compressor summary: The authors provide the first generalization bound for EGNNs, showing how spectral norms and layer weights affect their performance and propose a new regularizer based on these insights.
https://openreview.net/forum?id=YqMOM5W9GF
Compressor summary: We propose an optimistic algorithm for preferential Bayesian optimization using preference feedback and a confidence set, achieving an information-theoretic bound on cumulative regret and guaranteed convergence rate, outperforming existing heuristics without guarantees.
https://openreview.net/forum?id=YqIIhl2ToH
Compressor summary: The paper proposes a method to measure and quantify uncertainty in regression algorithms, helping detect unreliable behavior and improve error detection.
https://openreview.net/forum?id=YoUb2vW9WP
Compressor summary: The paper proposes new methods for selecting diverse subsets of categorical data using combinatorial optimization and submodular optimization techniques.
https://openreview.net/forum?id=YnFuUX08CE
Compressor summary: Round-trip correctness (RTC) is a new evaluation method for large language models that allows assessing their performance on various code-related tasks without manual curation, by checking if the model's prediction and synthesis match semantically.
https://openreview.net/forum?id=Yn8xnK90mS
Compressor summary: The paper studies how fast the EM algorithm converges for two-component mixed linear regression, providing explicit expressions and characterizing the trajectory of iterations, leading to improved error bounds and convergence exponent estimates.
https://openreview.net/forum?id=YlcSyCz21c
Compressor summary: The paper proposes a method called PaRe that uses a gating mechanism to transfer knowledge from large-scale pretrained models to various target modalities by creating intermediate data-rich modalities and improving cross-modal fine-tuning stability and performance.
https://openreview.net/forum?id=YlJy1FcM9E
Compressor summary: The paper analyzes the optimization landscape of supervised matrix factorization (SMF), derives applications, and proposes a block coordinate descent algorithm with convergence guarantees and a GPU-friendly neural implementation.
https://openreview.net/forum?id=YiblhkVl2w
Compressor summary: The paper studies ranking functions that account for predictors' uncertainty and shows they can achieve both stability and fairness in classification tasks.
https://openreview.net/forum?id=YdwwWRX20q
Compressor summary: FViTs are a rigorous approach to improve ViTs' explanations and robustness by applying Denoised Diffusion Smoothing (DDS) to their self-attention vectors and prediction distributions.
https://openreview.net/forum?id=Yd8eHMY1wz
Compressor summary: The text proposes Moirai, a new time series Transformer model that addresses challenges in universal forecasting by using a pre-trained Large Time Series Model on diverse datasets and achieves competitive or superior performance.
https://openreview.net/forum?id=YbHCqn4qF4
Compressor summary: The paper proposes a new efficient generic vision backbone, Vim, using bidirectional Mamba blocks that improve performance and memory efficiency over existing Transformer-based models on various computer vision tasks.
https://openreview.net/forum?id=YYwERRXsJW
Compressor summary: FedPGP is a method for federated prompt learning that balances generalization and personalization using CLIP guidance and low-rank adaptation.
https://openreview.net/forum?id=YWuSLBkfOw
Compressor summary: The paper proposes a small network (TempNet) to predict personalized temperature for large foundation models, which enhances their performance and can be applied to new tasks.
https://openreview.net/forum?id=YT1dtdLvSN
Compressor summary: OptiMUS is an LLM-based agent that can formulate and solve linear programming problems from natural language descriptions, outperforming existing methods on easy and hard datasets.
https://openreview.net/forum?id=YSoMmNWZZx
Compressor summary: RL-VLM-F automatically generates reward functions for agents to learn new tasks using text and visual inputs, by querying vision language models for preference feedbacks.
https://openreview.net/forum?id=YRWdiaupCr
Compressor summary: The paper proposes a novel Two-Stage Shadow Inclusion (2SSI) method that addresses latent confounding bias and collider bias in causal inference using the treatment residual as a shadow variable.
https://openreview.net/forum?id=YPbcUBcTAk
Compressor summary: PLCP is a novel framework that learns uncertainty-guided features for improving conditional validity of prediction sets using calibration data, machine learning models, and alternating gradient descent, with theoretical and empirical results showing its superior performance.
https://openreview.net/forum?id=YNvGFaOG1p
Compressor summary: The paper presents tight convergence bounds for actor-critic and natural actor-critic algorithms in reinforcement learning with compatible function approximation, addressing challenges such as stochastic bias and non-ergodicity.
https://openreview.net/forum?id=YNbCbcGyXE
Compressor summary: The study investigates how varying depth of transformer architecture affects its performance in different sequence learning tasks, showing that at least two attention layers are needed for reasoning and generalization, while three may be required for contextual generalization.
https://openreview.net/forum?id=YJWlUMW6YP
Compressor summary: The study warns that simplifying deep learning models may lead to inaccurate predictions when the models encounter out-of-distribution data, as they might overfit the training set.
https://openreview.net/forum?id=YEQM0asWCH
Compressor summary: CPR is a multi-task learning framework that enables interpretability and adaptability in complex decision processes by generating context-specific policies on-demand as observations change.
https://openreview.net/forum?id=YCzbfs2few
Compressor summary: IBD-PSC is a simple input-level backdoor detection method for deep neural networks that uses parameter-oriented scaling consistency to filter out malicious images and requires adaptive selection of batch normalization layers.
https://openreview.net/forum?id=YBetKvUlF7
Compressor summary: The paper studies how deep neural networks behave when trained on imbalanced datasets, showing that their features collapse to a structure of orthogonal vectors with lengths depending on the number of training samples.
https://openreview.net/forum?id=YBXwr7wF7i
Compressor summary: The paper analyzes the existence and uniqueness of fixed points for implicit-depth neural networks using subhomogeneous operators and Perron-Frobenius theory, leading to a more flexible framework for well-defined networks.
https://openreview.net/forum?id=YB1O99gK7b
Compressor summary: GEEX is a new method for explaining deep learning models without accessing their internal workings, which produces gradient-like explanations using only query-level access and has proven theoretical and empirical advantages over existing black-box methods.
https://openreview.net/forum?id=Y9qzwNlKVU
Compressor summary: RLE is a new exploration technique for deep RL that adds structured random rewards to original task rewards in random states to improve performance on Atari and IsaacGym benchmarks.
https://openreview.net/forum?id=Y8KsHT1kTV
Compressor summary: AD$^\varepsilon$ is a new method for in-context Reinforcement Learning that uses noise injection to learn from unlabeled data and improve performance without requiring optimal policies.
https://openreview.net/forum?id=Y5Zi59N265
Compressor summary: The text introduces GeoMFormer, a novel Transformer-based molecular model that learns invariant and equivariant features for molecular systems using cross-attention modules.
https://openreview.net/forum?id=Y5AmNYiyCQ
Compressor summary: The study introduces NLHF, an alternative pipeline for fine-tuning LLMs using pairwise human feedback that aims to align them better with human preferences.
https://openreview.net/forum?id=Y50K6DSrWo
Compressor summary: DiverseNO uses multiple neural operator heads to improve uncertainty estimates and out-of-domain performance, while Operator-ProbConserv updates the model with well-calibrated UQ estimates.
https://openreview.net/forum?id=Y4wxCICbD0
Compressor summary: The paper proposes Linear Alignment, a new algorithm that aligns AI assistants with human preferences in one step, without needing data annotation or model training, improving their performance and efficiency.
https://openreview.net/forum?id=Y4VgJfbjfl
Compressor summary: CuTS is a novel framework for generating customizable synthetic tabular data that supports various requirements and outperforms existing approaches in several tasks.
https://openreview.net/forum?id=Y2wRKE0Qor
Compressor summary: SINGD is a memory-efficient and numerically robust second-order method for neural net training that performs well even in low precision, addressing limitations of KFAC and outperforming AdamW.
https://openreview.net/forum?id=Y2WorV5ag6
Compressor summary: The paper proposes a new data-driven weather forecasting framework that combines an AI model with a traditional data assimilation algorithm to generate accurate and efficient global weather forecasts.
https://openreview.net/forum?id=Y0sH9HGMwq
Compressor summary: The paper studies how covariance information and uncertainty affects the accuracy of prediction in zero-sum games using deterministic learning dynamics and different discretization methods.
https://openreview.net/forum?id=XyxuhLtFA2
Compressor summary: The paper proposes an optimization-free method for slicing distribution selection that speeds up Monte Carlo estimation using random-path projecting direction and two variants of sliced Wasserstein.
https://openreview.net/forum?id=XyhgssAo5b
Compressor summary: RobustSL is a novel data structure that uses predictions of access frequencies to implement optimal and robust dictionaries with logarithmic runtime.
https://openreview.net/forum?id=XxCfToC9pJ
Compressor summary: The paper proposes Universal Entropy Optimization (UEO), a method that improves CLIP's performance on downstream tasks by adjusting textual prompts and visual transformations for out-of-distribution detection and known class recognition.
https://openreview.net/forum?id=XwnABAdH5y
Compressor summary: The paper proposes a reinforcement learning algorithm that helps language models become more trustworthy in retrieval augmentation without explicit supervision.
https://openreview.net/forum?id=XwVkqvyziD
Compressor summary: The paper analyzes stochastic weight averaging method's advantages over stochastic gradient descent in non-convex and realistic scenarios, using mathematical induction to derive stability and generalization bounds.
https://openreview.net/forum?id=XvmooikuHE
Compressor summary: The paper derives an exact expression for the probability distribution of hypervolume improvement in bi-objective optimization using Gaussian process models and proposes a new acquisition function, $\varepsilon$-probability of hypervolume improvement, which outperforms other methods in high uncertainty settings.
https://openreview.net/forum?id=XuQPA4D396
Compressor summary: PPS-VAE is a model that learns which pixels to observe in order to improve contextual image completion and extract meaningful information from images.
https://openreview.net/forum?id=XtDJaSe8jE
Compressor summary: AFT adaptively transfers useful pre-trained features to small, task-specific downstream models, improving performance and allowing combination of multiple pre-trained models.
https://openreview.net/forum?id=XsDWw1Mn2p
Compressor summary: The paper discusses how input space reconstruction can lead to misalignment between learning to reconstruct and learning for perception, and suggests that learning by denoising might be a better approach.
https://openreview.net/forum?id=XoencoHWy7
Compressor summary: CLIF is a method to improve text-guided diffusion models for generating customized concepts by fine-tuning CLIP using contrastive learning.
https://openreview.net/forum?id=XobPpcN4yZ
Compressor summary: VEB-RL combines evolutionary algorithms with value-based reinforcement learning, enhancing sample efficiency and performance for RL optimization.
https://openreview.net/forum?id=XoSF46Pc2e
Compressor summary: The text describes a framework to design differentially private algorithms that efficiently find approximate solutions for various classes of non-convex loss functions.
https://openreview.net/forum?id=XnsI1HKAKC
Compressor summary: PseudoCal is a novel post-hoc calibration framework for unsupervised domain adaptation that uses inference-stage mixup to generate labeled pseudo-target data, addressing the challenge of poorly calibrated predictive uncertainty on target data.
https://openreview.net/forum?id=XmLNDlQuzO
Compressor summary: Marginalization models are a new type of generative model that can quickly and flexibly approximate arbitrary marginal probabilities for high-dimensional discrete data, using energy-based training to enable any-order generative modeling.
https://openreview.net/forum?id=XlgeQ47Ra9
Compressor summary: The paper proposes a graph neural network to solve online Bayesian bipartite matching problems, which estimate expected weights of matchings and leverage local graph structures.
https://openreview.net/forum?id=XkHJo8iXGQ
Compressor summary: The paper explores the limitations of Instruction Tuning (IT) for large language models, showing that it fails to enhance knowledge or skills and can even degrade response quality.
https://openreview.net/forum?id=XiemSZpvh0
Compressor summary: The paper introduces Neuro-Visualizer, a new auto-encoder-based method for visualizing neural network loss landscapes that is more flexible and accurate than existing linear methods and provides useful insights for knowledge-guided machine learning applications.
https://openreview.net/forum?id=XhH1OKLANY
Compressor summary: LeaPformer is a new approach that uses dynamic proportions instead of fixed positions to improve linearized transformers' performance in various tasks.
https://openreview.net/forum?id=Xgrey8uQhr
Compressor summary: The paper proposes a novel data augmentation method for graph out-of-distribution generalization that extrapolates structure spaces to generate unseen graph data without compromising causal mechanisms.
https://openreview.net/forum?id=Xeh8171Fce
Compressor summary: The study explores how random feature models and Transformers learn in different domain settings and shows that minimal degree interpolators are only learned in specific cases like the Boolean setting with roots of unities.
https://openreview.net/forum?id=XecUTmB9yD
Compressor summary: FedCal is a novel approach for local and global calibration in federated learning that improves prediction accuracy and reduces calibration error by leveraging client-specific scalers and weight averaging.
https://openreview.net/forum?id=Xe7n2ZqpBP
Compressor summary: The paper discusses the challenges of conducting reliable reinforcement learning experiments and suggests an alternative approach due to high computational costs.
https://openreview.net/forum?id=Xdy9bjwHDu
Compressor summary: Shuffling gradient methods, a popular machine learning technique, have improved theoretical guarantees for their performance in different optimization settings.
https://openreview.net/forum?id=Xb3IXEBYuw
Compressor summary: The paper proposes a new algorithmic framework for solving bilevel RL problems with dynamic objective functions using penalty formulation and shows its effectiveness through simulations.
https://openreview.net/forum?id=XXioxiADDC
Compressor summary: The authors propose a new hypothesis that image data is distributed in "manifolds with skeletons" to explain why diffusion models struggle to generate detailed shapes like human hands, and they support their hypothesis through visualization and testing on natural images.
https://openreview.net/forum?id=XWkRyIjYDp
Compressor summary: The paper analyzes the stability and generalization of stochastic compositional optimization algorithms using statistical learning theory, introducing a new concept called "compositional uniform stability" and deriving dimension-independent excess risk bounds for two popular algorithms.
https://openreview.net/forum?id=XUeoOBid3x
Compressor summary: Magicoder is an open-source LLM for code that uses OSS-Instruct to create realistic and controllable synthetic instruction data, outperforming other models on various coding benchmarks.
https://openreview.net/forum?id=XUc29ydmLX
Compressor summary: The text discusses how Independent Subnetwork Training (IST) addresses communication and memory issues in large-scale machine learning by analyzing its optimization performance on a quadratic model.
https://openreview.net/forum?id=XUOHKSsurt
Compressor summary: FourierFT is a method that compresses trainable parameters in fine-tuning foundation models using the Fourier transform, achieving comparable or better performance than LoRA with fewer parameters.
https://openreview.net/forum?id=XTrMY9sHKF
Compressor summary: HarmonicFlow is a simple and general method to generate 3D protein-ligand binding structures, while FlowSite improves it further by jointly designing protein pockets and ligand structures for better binding site design.
https://openreview.net/forum?id=XTglHJjzQI
Compressor summary: Clifford-Steerable Convolutional Neural Networks (CS-CNNs) are a new type of equivariant CNNs that work with multivector fields and outperform baseline methods in fluid dynamics and relativistic electrodynamics.
https://openreview.net/forum?id=XT6iF8FDZx
Compressor summary: The paper investigates how policy gradient in reinforcement learning biases controllers towards unseen initial states based on exploration during training, and suggests selecting initial states for training to improve real-world performance.
https://openreview.net/forum?id=XSsoggg8pz
Compressor summary: The approach trains classifiers using available data and explores subpopulations to collect outcome data, ensuring fairness and convergence while minimally reducing predictive accuracy.
https://openreview.net/forum?id=XQz7ytgETQ
Compressor summary: The text introduces a new community detection method (TCD) that identifies tight communities without including scattered nodes, which improves accuracy and reveals biological implications in networks.
https://openreview.net/forum?id=XPP6K57bop
Compressor summary: The paper proposes a stability evaluation method using optimal transport that can handle data corruptions and sub-population shifts in real-world scenarios, and provides convex formulations and computational methods for different loss functions.
https://openreview.net/forum?id=XMlUlY7ONf
Compressor summary: MI allows us to understand how neural networks learn meaningful low-dimensional representations of complex data, which can reveal insights about the underlying problem and help derive new knowledge.
https://openreview.net/forum?id=XLlQb24X2o
Compressor summary: The paper proposes a test-time adaptation framework for open-set image restoration that adapts to unknown degradations using a diffusion model and an adapter.
https://openreview.net/forum?id=XKxuTZRCXq
Compressor summary: The paper proposes using differentially-private synthetic data to train ML models without revealing sensitive information from real data, and provides bounds on the risk of linear models trained on synthetic data.
https://openreview.net/forum?id=XGq30hC5MW
Compressor summary: The paper proposes a novel risk-sensitive exploration framework for reinforcement learning based on Conditional Value-at-Risk (CVaR), which works for any reward function and has near-optimal sample complexity.
https://openreview.net/forum?id=XGGcnKelda
Compressor summary: The paper proposes a new algorithm for post-processing ML predictions on RNA secondary structures that ensures biological relevance and improves performance.
https://openreview.net/forum?id=XDz9leJ9iK
Compressor summary: The paper discusses how foundation models could worsen existing disparities against marginalized communities in various aspects of AI and suggests ways to address these issues.
https://openreview.net/forum?id=XBNhJQU84y
Compressor summary: The paper proposes a method to fix topological artifacts in 3D shape generative models by using topological regularization losses on an implicit shape generator.
https://openreview.net/forum?id=X9VMhfFxwn
Compressor summary: The paper shows that using Soft MoE modules in value-based networks improves the scalability and performance of reinforcement learning models.
https://openreview.net/forum?id=X8uQ1TslUc
Compressor summary: The authors analyze CLIP using Optimal Transport and propose new losses for image and text tasks, as well as a graph-based inference method for zero-shot classification and related subtasks.
https://openreview.net/forum?id=X8Ha2NiQcy
Compressor summary: Sparse Iso-FOP Transformations (Sparse-IFT) improves accuracy of dense neural network models by efficiently using sparsity to maintain FLOPs.
https://openreview.net/forum?id=X7UnDevHOM
Compressor summary: Key points: - Pre-training improves efficiency and performance of neural operators for PDE data - New auto-regressive denoising pre-training strategy is stable, efficient, and generalizes well - Flexible and scalable model architecture based on Fourier attention allows large-scale pre-training - Achieve SOTA on 10+ PDE benchmarks and enhance performance on diverse downstream tasks Summary: The paper proposes a new pre-training strategy and a flexible model architecture for neural operators on PDE data, achieving state-of-the-art results and improving generalization to various tasks.
https://openreview.net/forum?id=WzD4a5ufN8
Compressor summary: This paper proposes new geometric representations and features for graph neural network-based computational fluid dynamics simulations, improving accuracy with low-resolution data.
https://openreview.net/forum?id=Wz4lgc8dsN
Compressor summary: Online cascade learning uses smaller models to imitate large language models, reducing inference costs while maintaining accuracy in data stream processing tasks.
https://openreview.net/forum?id=WwLtwPHmSM
Compressor summary: The paper studies sequential decision-making problems with improving options and proposes two algorithms to identify the best option within a fixed budget.
https://openreview.net/forum?id=WvvkbWD1vL
Compressor summary: The authors propose MoMo and MoMo-Adam, adaptive learning rates that improve performance and reduce tuning of momentum-based methods for various machine learning tasks.
https://openreview.net/forum?id=WvIHbQhrTq
Compressor summary: The paper proposes MIN-UCB, an online policy that uses offline data effectively when they are informative, and performs better than UCB in stochastic multi-armed bandits.
https://openreview.net/forum?id=WtvI3QijEF
Compressor summary: The paper analyzes how cognitive and expressive abilities in bilingual language models evolve during different phases of training, finding that cognitive capacity may limit expressive potential.
https://openreview.net/forum?id=WsawczEqO6
Compressor summary: Our study questions the connection between In-Context Learning (ICL) in pre-trained language models and Gradient Descent (GD), revealing inconsistencies in their behavior and output modifications, leaving the equivalence open.
https://openreview.net/forum?id=WsM4TVsZpJ
Compressor summary: The authors present a hierarchical RL framework that improves Decision Transformer by enabling seamless stitching of sub-optimal trajectories, leading to better offline RL performance on control and navigation tasks.
https://openreview.net/forum?id=WpKDeixmFr
Compressor summary: TIME WEAVER is a novel model that uses heterogeneous metadata to improve time series generation and introduces a new evaluation metric for conditional generation approaches.
https://openreview.net/forum?id=Wp054bnPq9
Compressor summary: The text discusses the vulnerabilities of language model watermarking schemes, which can be bypassed by watermark stealing attacks that enable spoofing and scrubbing of AI-generated content.
https://openreview.net/forum?id=WofwaWjIf7
Compressor summary: CROW is a prototype-based method that discovers and matches novel classes in cross-domain open-world settings using foundation models and a well-structured representation space.
https://openreview.net/forum?id=Wnni3cu39x
Compressor summary: This paper proposes a method to bound causal effects using instrumental variables under weak confounding and a criterion to falsify the IV with extra information on the confounder.
https://openreview.net/forum?id=Wnhp34K5jR
Compressor summary: The paper introduces universal gradient methods for stochastic convex optimization that adapt to noise and smoothness without prior knowledge and achieve state-of-the-art convergence rates.
https://openreview.net/forum?id=Wn4QwCrDvH
Compressor summary: CANVI is a scalable and easily implemented method for simulating posterior approximations with guaranteed marginal coverage and high predictive efficiency in likelihood-free settings.
https://openreview.net/forum?id=WjvEvYTy3w
Compressor summary: HyperDistill combines morphology-conditioned hypernetworks and policy distillation to learn efficient robot policies that perform as well as a universal transformer teacher on various morphologies, reducing model size and computational cost significantly.
https://openreview.net/forum?id=Wjq2bS7fTK
Compressor summary: FedREDefense is a method to defend against model poisoning attacks in federated learning by identifying and filtering out malicious clients based on discrepancies in their model update reconstruction errors.
https://openreview.net/forum?id=WjNzXeiOSL
Compressor summary: Key points: - SSM is a foundation model in time series analysis, alternative to transformers - Paper studies generalization of SSMs and proposes improvements to training algorithms - Data-dependent generalization bound for SSMs shows interplay between parameters and temporal dependencies - Scaling rule for initialization and new regularization method introduced based on generalization bound - Numerical results validate the proposed methods Summary: The paper analyzes the generalization of SSMs, a model in time series analysis, and proposes new training algorithms based on a data-dependent generalization bound that considers parameters and temporal dependencies.
https://openreview.net/forum?id=Wj5wm3Os5v
Compressor summary: QUAG probes reveal that VideoQA Transformers may not effectively use multimodal information and CLAVI dataset challenges them further.
https://openreview.net/forum?id=WfJuiIiFzB
Compressor summary: The paper proposes a dynamic labeling strategy for semi-supervised learning that uses shared and specific information to improve sample classification confidence and performance.
https://openreview.net/forum?id=WajJf47TUi
Compressor summary: This study introduces a generalizable machine learning architecture using graph neural networks that incorporates physical laws and symmetries to solve partial differential equations more reliably.
https://openreview.net/forum?id=WYi3WKZjYe
Compressor summary: Audio Flamingo is a new audio language model that can understand diverse sounds, adapt to tasks quickly, and engage in multi-turn dialogues, achieving state-of-the-art results on various audio understanding tasks.
https://openreview.net/forum?id=WXg6MJo1FH
Compressor summary: RLHF uses human feedback to teach AI models human values, but its reward model often deteriorates after one epoch; a new algorithm called 'Iterative Data Smoothing' improves it by updating data with soft labels instead of hard ones.
https://openreview.net/forum?id=WWo9G5zyh0
Compressor summary: The paper presents GeoReasoner, a vision-language model enhanced with human inference knowledge that improves geo-localization performance significantly.
https://openreview.net/forum?id=WVORGH73Cg
Compressor summary: The paper studies when optimizing one performance metric leads to optimizing another, focusing on error probability minimization under different learning criteria, and shows that non-monotonic criteria can prevent collapse in some cases.
https://openreview.net/forum?id=WUicA0hOF9
Compressor summary: The authors propose an evaluation framework for predictive allocation systems in settings with hierarchical units and show that their efficacy is limited by between-unit inequality, intervention budget, and other factors.
https://openreview.net/forum?id=WUi1AqhKn5
Compressor summary: AdaPromptCL is a new method for continual learning that adapts to varying degrees of semantic shifts between tasks using assign-and-refine semantic grouping.
https://openreview.net/forum?id=WUdq1WFUPr
Compressor summary: Cascade-CLIP aligns multi-level visual features with text embeddings using independent decoders to improve zero-shot semantic segmentation performance.
https://openreview.net/forum?id=WUQ4YzIQt2
Compressor summary: The paper proposes a data selection method using k-means clustering and sensitivity sampling to efficiently train machine learning models with low error.
https://openreview.net/forum?id=WT4X3QYopC
Compressor summary: Grad-ECLIP is a method to explain how CLIP, a vision-language model, matches image-text pairs by producing heat maps that show the influence of image regions or words on the results.
https://openreview.net/forum?id=WSpPC1Jm0p
Compressor summary: The paper proposes FreeShap, a more robust and efficient method for instance attribution in foundational models, which can improve explainability and performance in various data-centric tasks.
https://openreview.net/forum?id=WSi4IiMaCx
Compressor summary: The paper proposes an efficient algorithm for robust sparse mean estimation that runs in subquadratic time using few samples.
https://openreview.net/forum?id=WRIn2HmtBS
Compressor summary: The Hourglass Diffusion Transformer (HDiT) is a high-resolution image-generative model that uses the Transformer architecture and trains efficiently without special techniques.
https://openreview.net/forum?id=WQbDS9RydY
Compressor summary: The paper proposes a curvature-based metric to measure memorization in deep neural networks and shows its effectiveness in detecting mislabeled samples and unique failures.
https://openreview.net/forum?id=WPt9HRmMrG
Compressor summary: Key points: - Semantic segmentation requires pixel-wise annotations, which are hard to obtain and error-prone - The paper proposes an active label correction (ALC) framework that uses correction queries to fix pseudo labels of pixels - The method is more annotator-friendly than standard ones and leverages foundation models for zero-shot predictions on superpixels - The method improves semantic segmentation and label correction performance on various datasets, including PASCAL Summary: The paper presents an ALC framework that uses annotator-friendly correction queries to fix pixel-wise annotations for semantic segmentation, leveraging foundation models and superpixels.
https://openreview.net/forum?id=WPfYVdJHPk
Compressor summary: The paper discusses potential threats to decentralized machine learning training, proposes a poisoning attack, and suggests a robust training framework with detection and efficiency mechanisms.
https://openreview.net/forum?id=WOa96EG26M
Compressor summary: Larger language models are more sensitive to noise in the test context, while smaller ones are more robust to noise due to different attention mechanisms during in-context learning.
https://openreview.net/forum?id=WLPhywf1si
Compressor summary: The paper proposes an unsupervised adversarial fine-tuning method to make the CLIP vision encoder more robust against attacks that could spread fake information or defraud users in multi-modal foundation models like LLaVA and OpenFlamingo.
https://openreview.net/forum?id=WLGWMDtj8L
Compressor summary: COREP is a new algorithm that addresses complex non-stationarity in reinforcement learning by implicitly tracing the causal origin of changes in the environment and learning a stable graph representation for the state, leading to improved policy learning.
https://openreview.net/forum?id=WJn1BAx9aj
Compressor summary: The text introduces two models for matching graphs with corrupt nodes and studies their detection and estimation properties.
https://openreview.net/forum?id=WJ5fJhwvCl
Compressor summary: The study introduces S-DQN and S-PPO, novel algorithms that improve clean rewards, empirical robustness, and robustness guarantee in deep reinforcement learning with randomized smoothing.
https://openreview.net/forum?id=WIbntm28cM
Compressor summary: The paper proposes a new way to understand the function of individual neurons in neural networks by explaining them as linear combinations of concepts and evaluating their explanations using simulations.
https://openreview.net/forum?id=WIaZFk02fI
Compressor summary: The authors propose a new dataset, BREC, to measure the realized expressiveness of Graph Neural Networks (GNNs) that surpass the 1-dimensional Weisfeiler-Lehman test and show the gap between theory and practice in GNN expressiveness.
https://openreview.net/forum?id=WFyolnFZOR
Compressor summary: TutorEval and TutorChat are new NLP datasets for training language models to assist in STEM education using long textbook chapters and dialogues.
https://openreview.net/forum?id=WDgV1BJEW0
Compressor summary: Graph Sparse Training (GST) is a new method that dynamically adjusts the sparsity of graphs to optimize performance on Graph Neural Networks (GNNs) while preserving topological and semantic information.
https://openreview.net/forum?id=WCwxFM7n5S
Compressor summary: The paper proposes MFTPL-P, an oracle-efficient algorithm for interactive imitation learning with provable guarantees and wide applicability, and Bootstrap-DAgger, a practical variant without extra samples.
https://openreview.net/forum?id=WCVC5wGZyz
Compressor summary: Example Gisting is a novel approach for selecting informative examples that improves In-Context Learning performance in Large Language Models, and can be used for new tasks without additional training.
https://openreview.net/forum?id=WC14xZIaC2
Compressor summary: HyBRiD is a novel method that extracts maximally informative and minimally redundant high-order relationships from fMRI data to improve phenotypic predictions in neuroscience.
https://openreview.net/forum?id=W9GaJUVLCT
Compressor summary: The paper explores using Fourier analysis as an inductive bias for score-based diffusion models, showing that frequency diffusion models outperform time diffusion models on real-world datasets.
https://openreview.net/forum?id=W97gFmrKe6
Compressor summary: The paper investigates an inhomogeneous Wigner spike model to study structured noise and finds an optimal spectral method for detecting signal-noise separation.
https://openreview.net/forum?id=W8hBNk1FhQ
Compressor summary: The paper proposes a new semi-adaptive and variance-adaptive confidence set for linear bandits, which improves exploration efficiency when the noise level is unknown.
https://openreview.net/forum?id=W7Vqx1Jvc2
Compressor summary: The paper critiques the current machine learning research on Timeseries Anomaly Detection, highlighting issues with evaluation metrics and methods, and advocating for improved benchmarking practices and simpler models.
https://openreview.net/forum?id=W4pB7VbzZI
Compressor summary: FlowMM is a pair of generative models for predicting and proposing stable crystal structures using Riemannian Flow Matching with extended symmetries and flexibility, achieving state-of-the-art performance and efficiency in comparison to other methods.
https://openreview.net/forum?id=W4mLp5KuKl
Compressor summary: Key points: - Generalization theory of multi-label learning is still in early stage - Novel vector-contraction inequalities are developed to derive generalization bounds - Bounds depend weakly on the number of labels and capture label correlations - Macro-Averaged AUC bound is derived and analyzed with class-imbalance Summary: The paper develops novel inequalities to analyze and improve the generalization of multi-label learning, considering label correlations and class-imbalance.
https://openreview.net/forum?id=VyoY3Wh9Wd
Compressor summary: FT-PFN is a fast and accurate surrogate for Freeze-thaw Bayesian optimization, achieving state-of-the-art results in deep learning hyperparameter tuning.
https://openreview.net/forum?id=VyfEv6EjKR
Compressor summary: The paper proposes a method to improve graph neural networks by generating multiple graphs from latent variables and estimating their distribution using EM and MCMC techniques, leading to better node classification.
https://openreview.net/forum?id=VyGo1S5A6d
Compressor summary: P4D is a tool that tests the reliability of safety mechanisms in diffusion models by finding problematic prompts that can bypass them.
https://openreview.net/forum?id=VxI0gInNlh
Compressor summary: The paper studies symmetric positive semi-definite low-rank matrix completion with entry-dependent sampling and proposes a tailor-designed initialization for gradient descent to achieve global optimality.
https://openreview.net/forum?id=Vw4Yar2fmW
Compressor summary: The authors propose self-consistency training, a machine learning method for predicting molecular Hamiltonians in density functional theory without labeled data, which improves generalization and efficiency in data-scarce and out-of-distribution scenarios.
https://openreview.net/forum?id=VuoB86HiCL
Compressor summary: The paper proposes methods to estimate causal effects using proxy variables of unobserved confounding in linear models with multiple unmeasured confounders, without prior knowledge of their validity.
https://openreview.net/forum?id=VtqyurB4Af
Compressor summary: Diffusion with Spherical Gaussian constraint (DSG) is a method that improves conditional diffusion models by constraining guidance steps within data manifold, leading to better sample quality and faster sampling process.
https://openreview.net/forum?id=VsvfSMI5bs
Compressor summary: BAGEL is a method that helps language model agents learn to follow natural language instructions in digital environments by converting random trajectories into synthetic demonstrations using two noisy components, improving their performance and reducing failures.
https://openreview.net/forum?id=VrwIrAa1Lc
Compressor summary: Random Masking is a simple and effective method to fine-tune language models using fewer parameters by creating a flatter loss landscape and allowing larger learning rates.
https://openreview.net/forum?id=VoMPNYTZud
Compressor summary: SSL models use less labeled data and are hard to evaluate with validation sets, so a new method called SLAM combines spectral complexity and margin distribution to improve their generalization performance without relying on validation data.
https://openreview.net/forum?id=VnI9200eeL
Compressor summary: Auctionformer is an efficient transformer-based method that solves equilibria of diverse auctions using tokenization and Nash error as a loss term, outperforming existing approaches.
https://openreview.net/forum?id=ViZcgDQjyG
Compressor summary: The paper explores the potential limitations of large language models due to the limited availability of public human-generated text data and suggests alternative methods to continue improving them.
https://openreview.net/forum?id=VfWrXJtLSL
Compressor summary: The paper studies pure private learning in the agnostic model, deriving improved upper bounds for item- and user-level privacy and presenting tighter results for user-level privacy and learning thresholds.
https://openreview.net/forum?id=Vdr87ZUfnl
Compressor summary: The paper analyzes the regret of reinforcement learning algorithms for an admission control problem in queuing systems and proposes a new algorithm with improved bounds.
https://openreview.net/forum?id=VdZfEMuoj2
Compressor summary: The proposed method generates a tailored search space for Neural Architecture Search using a two-level generative model hierarchy, achieving state-of-the-art performance with low computational costs on various tasks.
https://openreview.net/forum?id=VaZVZQSgTP
Compressor summary: The paper proposes a new single-loop robust policy gradient method for solving Markov Decision Processes with global optimality guarantee and better convergence performance than existing methods.
https://openreview.net/forum?id=Va7mhTVy5s
Compressor summary: VoroNav is a novel semantic exploration framework for household robots that uses Reduced Voronoi Graph to generate text-based descriptions of paths, enabling commonsense reasoning and outperforming existing methods in ZSON task.
https://openreview.net/forum?id=VZsxhPpu9T
Compressor summary: The paper proposes a Rényi divergence-based variant of Pufferfish to improve its applicability, utility, and composition guarantees for privacy preservation in machine learning algorithms.
https://openreview.net/forum?id=VZ5A0LPbnc
Compressor summary: The paper proposes codebook features, which quantize continuous neural network features into discrete vector codes, enabling interpretability and control over the model's behavior.
https://openreview.net/forum?id=VWCpm39peL
Compressor summary: The paper presents a novel framework called Discrete Latent Perspective Learning (DLPL), which enables networks to understand images from different perspectives using single-view images and improves performance on various computer vision tasks.
https://openreview.net/forum?id=VW7Jk8KhNC
Compressor summary: The paper introduces an online algorithm for detecting changes in data from Riemannian manifolds using a generalized Karcher mean computed by stochastic Riemannian optimization, and provides theoretical and empirical performance analysis.
https://openreview.net/forum?id=VUTyzH63Xa
Compressor summary: Label noise SGD makes two-layer neural networks simple by always using a single linear feature across all neurons.
https://openreview.net/forum?id=VSwrXRqD9o
Compressor summary: The authors propose a method to solve dynamic programming problems with variational Bayesian inference, Gibbs distributions, and message-passing algorithms, and apply it to generate text-to-speech and singing voice.
https://openreview.net/forum?id=VRv8KjJNuj
Compressor summary: EquiCSP is a new equivariant diffusion-based model for Crystal Structure Prediction (CSP) that improves accuracy and convergence compared to existing methods.
https://openreview.net/forum?id=VOcsmIBiXE
Compressor summary: PiT is a more efficient and interpretable attention mechanism for Transformer-based operator learning, inspired by numerical methods for PDEs.
https://openreview.net/forum?id=VJwsDwuiuH
Compressor summary: The paper presents a fast and efficient policy optimization algorithm that achieves the optimal regret bound in both stochastic and adversarial online Markov decision processes.
https://openreview.net/forum?id=VJjjNrUi8j
Compressor summary: The text discusses challenges in measuring predictive uncertainty using second-order probability distributions and proposes formal criteria and a framework for developing better uncertainty measures, using the Wasserstein distance as an example.
https://openreview.net/forum?id=VHtIDVaOKC
Compressor summary: The paper proposes Mobile-Attention, a novel attention mechanism for ViTs that balances efficiency and capability on mobile devices by using a head-competition mechanism and information flow to prevent overemphasis on less important subspaces while preserving essential ones.
https://openreview.net/forum?id=VHO4nE7v41
Compressor summary: MeZO-SVRG is a memory-efficient optimization method for fine-tuning language models, improving accuracy and reducing computation time compared to previous methods.
https://openreview.net/forum?id=VF177x7Syw
Compressor summary: The study proposes Hare \& Tortoise, a method for neural networks that combines rapid adaptation and gradual knowledge integration to balance plasticity and generalization.
https://openreview.net/forum?id=VE3yWXt3KB
Compressor summary: The text describes a method to extract information from black-box language models by exploiting their API access, revealing hidden dimensions and projection matrices.
https://openreview.net/forum?id=VDgfJnOEMV
Compressor summary: The paper proposes a general convergence rate guarantee for BW gradient descent with constraints and provides a fast implementation method for constrained problems, showing improved performance in experiments.
https://openreview.net/forum?id=VAKkoJjVpn
Compressor summary: The authors propose a novel neural network preconditioner for Poisson equations with mixed boundary conditions that works efficiently even when domain shapes, boundary conditions, or grid sizes change and outperforms existing methods.
https://openreview.net/forum?id=V4qV08Vk6S
Compressor summary: LEO is a multi-modal agent that excels in various 3D tasks, trained with 3D vision-language and -action alignment using large-scale datasets and LLMs.
https://openreview.net/forum?id=V3ya8RlbrW
Compressor summary: The authors provide theoretical insights into the performance of contrastive continual learning, propose a novel algorithm called CILA that uses adaptive distillation coefficients, and achieve state-of-the-art results on standard benchmarks.
https://openreview.net/forum?id=V3OpGwo68Z
Compressor summary: Flash-Diffusion is a novel method that adapts compute power and inference times to the difficulty of reconstruction tasks in inverse problems using estimated degradation severities.
https://openreview.net/forum?id=Uzb45nolTb
Compressor summary: This text discusses how lottery tickets in deep learning are sparse subnetworks that require heavy data-dependent masks and pruning near initialization is infeasible due to mutual information between sparsity mask and data.
https://openreview.net/forum?id=Uz4Qr40Y3C
Compressor summary: Connect Later framework combines pretraining with targeted augmentations to improve out-of-distribution generalization for domain adaptation tasks.
https://openreview.net/forum?id=UqoG0YRfQx
Compressor summary: EWoK is an online approach that learns robust policies for sequential decision-making by simulating the worst transition kernel scenarios while using any off-the-shelf non-robust RL algorithm.
https://openreview.net/forum?id=UpSe7ag34v
Compressor summary: The study investigates time directionality in Autoregressive LLMs and finds a consistent but subtle difference in their predictability, which they explain using sparsity and computational complexity.
https://openreview.net/forum?id=Uoved2xD81
Compressor summary: Key points: - The paper proposes Action-conditioned World Models (AWMs) for reinforcement learning with transformers - AWMs provide more direct routes for gradient propagation and easier optimization landscapes - AWMs outperform baselines in long-horizon tasks Summary: The paper introduces AWMs, a class of world models that use transformers and condition on actions to improve gradient propagation and policy learning in long-horizon reinforcement learning problems.
https://openreview.net/forum?id=Uo3LNg5SLY
Compressor summary: FDTempExplainer is a novel explanation method for black-box time series models that reveals temporal interactions and outperforms existing approaches.
https://openreview.net/forum?id=UjDp4Wkq2V
Compressor summary: The paper shows that, for non-polynomial activation functions, MPNNs with constant-dimensional feature vectors can simulate the WL test, by using linear independence over rationals instead of reals.
https://openreview.net/forum?id=Uh5XN9d2J4
Compressor summary: This paper proposes reconstruction granularity as a novel solution to mitigate outliers in quantized transformer models and develops an algorithm for finding the optimal granularity, achieving state-of-the-art performance in post-training quantization.
https://openreview.net/forum?id=Ug1m4P4AKf
Compressor summary: The paper presents a new method for modeling multiple speakers, improving their performance on feature learning and representation, enabling subjective similarity evaluation and generation of artificial speakers.
https://openreview.net/forum?id=UdXDUDxq11
Compressor summary: The paper introduces a new reinforcement learning approach based on differential game theory that improves robustness to uncertainty and outperforms existing methods in real-world applications.
https://openreview.net/forum?id=UcOze9EXEc
Compressor summary: DFML can adapt to new tasks without data by using pre-trained models, but it needs to consider model heterogeneity and balance task conflicts and overfitting risk for effective learning.
https://openreview.net/forum?id=UZstTlLq1E
Compressor summary: The paper analyzes how the structure of a deep neural network can lead to all predictions being assigned to the same class before training, influenced by factors like preprocessing methods, activation functions, and network depth.
https://openreview.net/forum?id=UZlMXUGI6e
Compressor summary: This study introduces t-PatchGNN, a novel method to model correlations between irregular multivariate time series using transformable patches and adaptive graphs for improved forecasting.
https://openreview.net/forum?id=UZZaWUR0n4
Compressor summary: The text discusses a novel algorithm that efficiently ranks items/players with varying values using optimal matching and minimizing comparison costs.
https://openreview.net/forum?id=UW5nO9NGjt
Compressor summary: The paper proposes and analyzes Momentum Knowledge Distillation (MKD) for Online Continual Learning (OCL), a challenging problem where neural networks are trained on a continuous data stream with severe constraints, and shows its effectiveness in improving accuracy.
https://openreview.net/forum?id=UTSCK582Yo
Compressor summary: The Graph Positional and Structural Encoder (GPSE) is a novel graph encoder that captures rich positional and structural information for augmenting any GNN, improving their performance across various tasks and datasets.
https://openreview.net/forum?id=URtUYfC3GA
Compressor summary: WAVES is a benchmark for evaluating image watermark robustness against various attacks, exposing vulnerabilities in current algorithms.
https://openreview.net/forum?id=UQYXZdca92
Compressor summary: The paper proposes a method to generate probabilistic forecasts of dynamical systems using generative models and stochastic interpolants, and applies it to various problems like fluid dynamics and video prediction.
https://openreview.net/forum?id=ULleq1Dtaw
Compressor summary: The paper proposes a novel framework that learns interpretable branching policies for exact combinatorial optimization solvers using graph symbolic discovery and Transformer models, outperforming existing methods.
https://openreview.net/forum?id=ULKvSqmSgA
Compressor summary: The paper proposes an online learning algorithm for MLR with multiple sub-models and arbitrary mixing weights, which can converge without i.i.d. input data and achieve good data clustering performance.
https://openreview.net/forum?id=UKHfmzLR7P
Compressor summary: The paper proposes a variant of Collaborative PAC Learning that can learn accurate classifiers for each data distribution using fewer samples, under a weaker realizability assumption.
https://openreview.net/forum?id=UIxOkdBmxh
Compressor summary: Bayesian Persuasion uses machine learning to predict popularity and misinformation features of posts, then strategically signals this advantage to users to prevent them from sharing false information on social media platforms.
https://openreview.net/forum?id=UGpGkLzwpP
Compressor summary: The paper formalizes the "linear representation hypothesis" using counterfactuals and introduces a causal inner product that unifies geometric notions in concept representation space.
https://openreview.net/forum?id=UE79AkNg60
Compressor summary: The paper explores the relationships between different types of neural network components (HyperNetworks, weight prediction, gating, and kernel prediction) and shows how they can be combined for image denoising.
https://openreview.net/forum?id=UCKFhc9SFC
Compressor summary: Volume-MCTS is a tree search algorithm for robot navigation that combines policy optimization with state-occupancy measure regularization, leading to improved long-horizon exploration compared to AlphaZero.
https://openreview.net/forum?id=U97MIrs35l
Compressor summary: The paper proposes using morph-tokens to handle visual input for both textual comprehension and image generation in multimodal LLMs, achieving state-of-the-art performance in both tasks.
https://openreview.net/forum?id=U841CrDUx9
Compressor summary: The text proposes a general algorithm for various decision-making problems and presents three contributions, including a generalized mirror descent, a configurable mirror descent with a meta-controller, and a GameBench with 15 academic-friendly games.
https://openreview.net/forum?id=U4Yvwu1RQY
Compressor summary: The paper introduces exponential spectral pursuit (ESP), a new method for initializing sparse phase retrieval that has better sampling complexity and performance than current methods.
https://openreview.net/forum?id=U1uKihiG39
Compressor summary: This paper studies how to adapt to causal structure in multi-armed bandits, achieving better performance when possible and ensuring worst-case guarantees, while addressing open questions and assumptions in the field.
https://openreview.net/forum?id=TzqmqZS0nj
Compressor summary: The text discusses how AI machines use probabilistic machine learning to make decisions based on true facts in their environment and investigates conditions for learning these probabilities.
https://openreview.net/forum?id=TwZ2sY6eJj
Compressor summary: The paper proposes LatentTracer, a method to trace images generated by latent generative models without extra steps during training or generation, based on gradient-based latent inversion and encoder-based initialization.
https://openreview.net/forum?id=TvoG41N1Y3
Compressor summary: The paper presents a new machine learning method for predicting electron density in chemical systems using plane-wave and Gaussian-type orbitals, outperforming existing methods.
https://openreview.net/forum?id=TujtZgdRxB
Compressor summary: The paper explores using adversarial reinforcement learning to improve algorithms for an online bipartite matching problem, and discovers structural properties that reduce the hardness of the problem.
https://openreview.net/forum?id=TuALw8xVum
Compressor summary: The text proposes a new regularization method for training GANs with limited data, based on renormalization group ideas, which improves their performance and stability.
https://openreview.net/forum?id=TtSFg4s3F0
Compressor summary: The text discusses the need for efficient privacy protection in deep neural networks using a logistic mechanism instead of Gaussian noise.
https://openreview.net/forum?id=TqcZfMZjgM
Compressor summary: PinNet is a framework for generating high-quality code descriptions using retrieval augmentation, a discriminator to measure the relevance of retrieved references, and contrastive learning to enhance attention weights.
https://openreview.net/forum?id=ToHkAg936Y
Compressor summary: ClawNOs are neural operators that automatically satisfy fundamental conservation laws for various scientific applications, improving learning efficiency and physical consistency.
https://openreview.net/forum?id=Th8JPEmH4z
Compressor summary: The authors propose a framework that integrates large language models with symbolic reasoning components for better planning and reasoning tasks.
https://openreview.net/forum?id=TfwGtfPkhV
Compressor summary: The paper studies how to test if a linear bandit problem is feasible using low-regret algorithms and provides reliable and efficient tests with minimax lower bounds.
https://openreview.net/forum?id=TfWKkSAziC
Compressor summary: LPGD is a framework for training machine learning architectures with embedded optimization layers, which computes meaningful replacements of degenerate derivatives by re-running the forward solver on perturbed input and converges faster than gradient descent.
https://openreview.net/forum?id=TejqrQBvll
Compressor summary: This paper proposes a causal machine learning theory with generalization bounds, using a new inequality to measure treatment propensities and handling hidden confounding and non-positive data.
https://openreview.net/forum?id=TaAqeo7lUh
Compressor summary: Key points: - Study how to scale language models' context lengths to 128K with continual pretraining on data mixture - Hypothesize that long context modeling is mostly achieved through large-scale pretraining and can be extended with lightweight continual pretraining - Investigate the quantity and quality of data for continual pretraining, emphasizing domain balance and length upsampling - Show that 1B-5B tokens are enough for retrieving information anywhere within 128K context - Recipe outperforms open-source long-context models and approaches GPT-4 128K Summary: The paper proposes a continual pretraining recipe to scale language models' context lengths to 128K using data mixture, with a focus on data quality and quantity. The method uses 1B-5B tokens and achieves competitive performance to GPT-4.
https://openreview.net/forum?id=TWu1fzFJm0
Compressor summary: The text discusses how adjusting the learning rate in deep neural networks can enable feature learning and improve non-linear function recognition.
https://openreview.net/forum?id=TUKOklS3gg
Compressor summary: The choice of weights in pseudo-outcome regressions is more important than the transformation for estimating conditional average treatment effects.
https://openreview.net/forum?id=TTZXl9WYFF
Compressor summary: The text describes a novel reinforcement learning approach for optimizing emergency responder management that significantly reduces computation time and improves ambulance response times.
https://openreview.net/forum?id=TTYVG17wfc
Compressor summary: The paper introduces OSN, a framework to learn all plausible 3D scene configurations from a monocular RGB video using an object scale network and a joint optimization module.
https://openreview.net/forum?id=TRrXkVdhwi
Compressor summary: CHELA combines short-long convolutions with hardware-efficient linear attention to achieve global abstraction, data-dependent selection, and linear computational complexity for long sequences.
https://openreview.net/forum?id=TN3fi7dwPo
Compressor summary: Tandem transformers combine a small autoregressive model and a large model in block mode to improve inference speed and accuracy for language models.
https://openreview.net/forum?id=TK7xkOsXDu
Compressor summary: Key points: - The paper introduces Hierarchical State-Space models (HiSS) for continuous sequential prediction from raw sensory data. - HiSS stacks structured state-space models to create a temporal hierarchy and outperforms existing sequence models. - HiSS is efficient on small datasets and compatible with data-filtering techniques. Summary: The paper proposes HiSS, a novel technique for predicting physical quantities from raw sensory data using a temporal hierarchy of structured state-space models that beats current methods and works well on limited data and filtering.
https://openreview.net/forum?id=TJCUrzhbiH
Compressor summary: Diff history is a method to simplify and focus textual inputs for neural language models in embodied control tasks, leading to improved performance and reduced training data requirements.
https://openreview.net/forum?id=TJ6tVNt6Y4
Compressor summary: The paper proposes two energy-based methods for detecting and removing backdoors in machine learning models with low overhead, called EBBA and EBBA+.
https://openreview.net/forum?id=THPjMr2r0S
Compressor summary: The paper proposes and benchmarks various zeroth-order optimization techniques for memory-efficient fine-tuning of large language models, revealing new principles and introducing novel enhancements.
https://openreview.net/forum?id=T0zR4mdSce
Compressor summary: PARCv2 is a versatile and generalizable deep learning model that can simulate complex physical systems involving unsteady, transient, and advection-dominated dynamics.
https://openreview.net/forum?id=T0lFfO8HaK
Compressor summary: The paper proposes a sparse inversion method for high-resolution images that selectively reconstructs semantic foregrounds, skipping noisy backgrounds and spurious correlations, to accelerate model inversion while maintaining or improving downstream performance.
https://openreview.net/forum?id=Sz9mAYuqlE
Compressor summary: Key points: - Cooperative learning scenario with networked agents updating predictions through communication or observations - Goal: optimize a few classifiers to maximize overall accuracy in the network - Objective functions: aggregate and egalitarian - Results: polynomial time algorithm for aggregate, NP-hard for egalitarian, approximation algorithms for improvement, guaranteed performance Summary: The paper studies how to improve the accuracy of cooperative learning agents by optimizing a few classifiers using different objective functions, and presents efficient algorithms with proven guarantees.
https://openreview.net/forum?id=SyY7ScNpGL
Compressor summary: The paper questions Transformers' effectiveness in partially observable environments and suggests using Deep Linear Recurrent Unit (LRU) instead.
https://openreview.net/forum?id=SvvvB5t5EW
Compressor summary: The paper analyses different contrastive learning methods and shows they all optimize for the same problem related to hyperspherical energy minimization; it introduces a new objective called Decoupled Hyperspherical Energy Loss (DHEL) that simplifies the problem and improves performance and robustness on computer vision tasks.
https://openreview.net/forum?id=SvBLKoBL4q
Compressor summary: The paper studies Monotonic Linear Interpolation (MLI) and shows that its error decrease is mainly due to the converged model's property, not the optimization trajectory. It also identifies different scale invariance properties of the converged model.
https://openreview.net/forum?id=Sv4u9PtvT5
Compressor summary: The paper introduces GoldE, a framework for knowledge graph embedding that can handle different dimensions and geometries using a generalized Householder reflection parameterization, leading to improved modeling capability and performance.
https://openreview.net/forum?id=Su0qe33cWA
Compressor summary: Wasserstein Wormhole is a neural network that embeds distributions into a space where Euclidean distances approximate Wasserstein distances, enabling fast and scalable analysis of non-Euclidean data.
https://openreview.net/forum?id=Stn8hXkpe6
Compressor summary: ReMax is a simpler and more efficient reinforcement learning algorithm for large language models that leverages their properties and outperforms PPO.
https://openreview.net/forum?id=Ss3h1ixJAU
Compressor summary: The paper introduces a new reinforcement learning objective function that guarantees monotonic improvement in the lower probability bound of performance and develops two practical solutions (APO and PAPO) that significantly outperform state-of-the-art algorithms on continuous control tasks and Atari games.
https://openreview.net/forum?id=Sra298VMFM
Compressor summary: The paper proposes a new method called VI-IGL that uses information theory to enforce conditional independence assumptions and improve feedback-based reinforcement learning tasks.
https://openreview.net/forum?id=SoqxSnEUi1
Compressor summary: Albatross is a new algorithm that learns to play simultaneous games by modeling other agents' behavior and can cooperate or compete with agents of any strength, achieving better results than AlphaZero in some cases.
https://openreview.net/forum?id=SoNexFx8qz
Compressor summary: The paper proposes using smaller generative models together instead of large monolithic ones, which leads to more data-efficient learning, generalization to unseen data, and the ability to create new models for unknown tasks.
https://openreview.net/forum?id=SlRcJvf1yd
Compressor summary: The paper analyzes different methods to compute derivatives of non-differentiable functions, which are useful for machine learning applications like hyperparameter optimization and data poisoning attacks.
https://openreview.net/forum?id=SkI6u81AkI
Compressor summary: This paper presents a framework for using spiking neural networks (SNNs), which mimic biological neurons, to improve time-series forecasting with less energy consumption and better capture of temporal dependencies.
https://openreview.net/forum?id=Sjv5RcqfuH
Compressor summary: GATE is an extension of Graph Attention Networks that improves over-smoothing by addressing its root cause, allows for higher depth and better feature transformations, and outperforms GATs on heterophilic datasets.
https://openreview.net/forum?id=ShkKSDrfG6
Compressor summary: OTMatch is a new semi-supervised learning method that uses optimal transport loss to match class distributions and improve learning performance by leveraging semantic relationships among classes.
https://openreview.net/forum?id=SfcB4cVvPz
Compressor summary: The text explains how backdoor poisoning attacks work, why they are effective, and provides analysis and experiments on a specific type of neural network.
https://openreview.net/forum?id=Sf5KYznS2G
Compressor summary: Reflected flow matching trains a velocity model for continuous normalizing flows with boundary constraints to produce natural and class-conditional samples on constrained domains.
https://openreview.net/forum?id=ScRhEuj480
Compressor summary: The paper proposes a simple and scalable corrector network that adjusts stale cached target embeddings for dense retrieval, achieving state-of-the-art results with significant reductions in computational cost.
https://openreview.net/forum?id=ScIHQoTUjT
Compressor summary: The authors propose an evaluation framework for assessing how well large language models communicate about climate change, revealing a significant gap between their surface-level and deeper knowledge.
https://openreview.net/forum?id=Sbl2keQEML
Compressor summary: The paper proposes a method called "Surgery" to reduce representation bias in multi-task learning by aligning the merged model's representation with individual models' representations using an unsupervised optimization objective.
https://openreview.net/forum?id=SZ0JnRxi0x
Compressor summary: The paper introduces a new algorithm for determining reference frames and normalizing 3D point clouds that works universally, guarantees compatibility, and outperforms existing methods.
https://openreview.net/forum?id=SXVn5IFsrs
Compressor summary: The paper introduces CodeIt, a novel method for language models to self-improve by iterating between program sampling, hindsight relabeling, and learning from prioritized experience replay, achieving state-of-the-art performance on the Abstraction and Reasoning Corpus.
https://openreview.net/forum?id=SWrwurHAeq
Compressor summary: SiT is a scalable vision transformer that uses Graph Symmetric Attention to improve generalisation in reinforcement learning by preserving graph symmetries and leveraging local and global data patterns.
https://openreview.net/forum?id=SUxarNgrUT
Compressor summary: The paper proposes a new adaptive method for convex problems that doesn't need Lipschitz assumptions and uses plain Hölder inequalities to achieve linesearch-free convergence without prior knowledge of constants or order.
https://openreview.net/forum?id=SRzb3QDjdV
Compressor summary: The paper proposes PIDformer, a new class of transformers that use feedback control to improve robustness, representation capacity, and noise resilience by avoiding input corruption and rank collapse issues in existing models.
https://openreview.net/forum?id=SRmZw7nEGW
Compressor summary: UniAudio is a large language model-based universal audio generation model that can handle various audio tasks using different inputs, such as phonemes, text descriptions, or existing audio, and achieve competitive results on 11 tasks.
https://openreview.net/forum?id=SQIDlJd3hN
Compressor summary: RoboGen is a robotic agent that learns skills from simulated environments using foundation and generative models, creating diverse tasks and supervisions with minimal human input.
https://openreview.net/forum?id=SPygKwms0X
Compressor summary: The paper proposes a new decision rule for out-of-distribution (OOD) detection based on a generalized Benjamini Hochberg procedure that has rigorous theoretical guarantees and performs well empirically.
https://openreview.net/forum?id=SPBxFwIdMk
Compressor summary: PRESTO is a framework that analyzes the variability in latent representations of machine learning models using persistent homology, enabling sensitivity analysis, anomaly detection, and hyperparameter search.
https://openreview.net/forum?id=SMUXPVKUBg
Compressor summary: FOIL is a model-agnostic framework that improves out-of-distribution generalization in time-series forecasting via invariant learning, addressing challenges such as unobserved core variables and lack of environment labels.
https://openreview.net/forum?id=SLqdDWwibH
Compressor summary: The text describes a new method that uses adversarial samples to improve the quality of Neural Signed Distance Functions (SDF) for 3D shape representation, outperforming existing methods on synthetic and real data.
https://openreview.net/forum?id=SL6V527p1F
Compressor summary: The paper proposes a novel identifiability condition for causal representation learning based on observational variable grouping, with a self-supervised estimation framework that outperforms existing methods.
https://openreview.net/forum?id=SKPhvzxO1g
Compressor summary: The paper proposes a framework called TRACE that explains feature deletion influence on model predictions and benchmarks insertion/deletion metrics for explanation methods.
https://openreview.net/forum?id=SERrqPDvoY
Compressor summary: The paper proposes a two-stage model for High-Resolution Salient Object Detection, which preserves details and is fast enough for real-time applications.
https://openreview.net/forum?id=SE20BFqj6J
Compressor summary: D-Flow is a method to control the output of diffusion and flow-matching models by differentiating through their generation process, achieving excellent results in various tasks such as image and audio inversion and conditional molecule generation.
https://openreview.net/forum?id=SDCx6rQV2l
Compressor summary: The paper proposes a new method for selective classification called CCL-SC that improves feature layers and reduces selective risk by contrasting similar and different instances based on confidence.
https://openreview.net/forum?id=SBR8Gwe1E2
Compressor summary: The paper introduces DetKDS, a framework that uses search algorithms to automatically find optimal detection distillation policies for different detector setups, achieving state-of-the-art results on various tasks.
https://openreview.net/forum?id=SAbZExIIgG
Compressor summary: We propose a federated optimization method for decomposable submodular functions that respects privacy and reduces communication cost by aggregating only intermittently and on a subsampled set of clients.
https://openreview.net/forum?id=SAbL40d8A4
Compressor summary: The paper proposes a new method to use Winner-takes-all learners for conditional density estimation that leverages their geometric properties and shows its advantages in theory and practice.
https://openreview.net/forum?id=SAXp5dMYv7
Compressor summary: This text describes a method to improve Markov chain Monte Carlo samplers by using bijective affine transformations and adaptive learning, which can generate high-quality samples efficiently for real-world data.
https://openreview.net/forum?id=SAEUO7847g
Compressor summary: The paper proposes LowPopArt, a novel method for low-rank matrix estimation in trace regression and bandits, with tighter recovery guarantees and improved regret bounds than existing methods.
https://openreview.net/forum?id=S9lk6dk4LL
Compressor summary: The paper proposes a method to pre-train LLMs on videos by decomposing them into keyframes and motions, and then adapting them to the model using tokenizers that discretize visual and temporal information, allowing for unified pre-training of images, videos, and text.
https://openreview.net/forum?id=S9DV6ZP4eE
Compressor summary: The adaptive analytic gradient method improves policy optimization using simulation gradients by adjusting the Q function to handle non-smooth simulations, and is demonstrated on AGPO algorithm with theoretical and empirical results showing its effectiveness.
https://openreview.net/forum?id=S80a4hJtuE
Compressor summary: The paper proposes an efficient algorithm for offline reinforcement learning with linear MDPs under infinite-horizon discounted setting and partial data coverage assumption, achieving better sample complexity than existing methods.
https://openreview.net/forum?id=S6a6gHvMWx
Compressor summary: The paper introduces Social Environment Design, a framework that uses AI to automate policy-making, connects with different research communities, and addresses key open problems for future research in this field.
https://openreview.net/forum?id=S4LqI6CcJ3
Compressor summary: BEYOND is a novel AE detection framework that uses SSL to distinguish between clean samples and AEs based on representation similarity and label consistency, achieving state-of-the-art robustness accuracy.
https://openreview.net/forum?id=S3xqyEaST9
Compressor summary: The paper proposes algorithms and methods for optimizing pipeline parallelism in deep neural network inference by minimizing the running time of the bottleneck stage and provides empirical results on a diverse testbed of production models.
https://openreview.net/forum?id=S2XgbBCJy0
Compressor summary: The paper studies how transaction costs can improve data markets by reducing negative externalities from buyers' purchases and proposes learning algorithms to achieve low regret in valuation.
https://openreview.net/forum?id=S1gSrruVd4
Compressor summary: The text argues against assuming conditionally independent probabilities in neurosymbolic learning systems, as it can cause overconfidence, poor uncertainty representation, and optimization difficulties.
https://openreview.net/forum?id=S0DPCE7tt4
Compressor summary: The text discusses the importance of 'shaping' procedures for complex tasks in systems neuroscience and proposes a model to analyse deep policy gradient learning of compositional reinforcement learning tasks using statistical physics tools.
https://openreview.net/forum?id=Ry4RAzdOWl
Compressor summary: Event-based vision method EvTexture uses high-frequency event details for texture enhancement in video super-resolution, achieving state-of-the-art performance.
https://openreview.net/forum?id=Rx9GMufByc
Compressor summary: The study introduces $ ext{aL\small{LM}4T\small{S}}$, a framework that adapts Large Language Models for time-series representation learning using self-supervised, multi-patch prediction.
https://openreview.net/forum?id=RvwMTDYTOb
Compressor summary: The paper presents a new algorithm (FSA) that makes support vector machines and quantile regression faster and more accurate for high-dimensional data by transforming their non-smooth loss functions into smooth ones.
https://openreview.net/forum?id=RuH78kOcDi
Compressor summary: Key points: - Decentralized bilevel optimization is useful for many domains but hard to ensure differential privacy. - The paper proposes a new algorithm that achieves both differential privacy and accurate convergence. - The paper also analyzes the convergence rate and the price of differential privacy. Summary: The paper presents a new decentralized bilevel optimization algorithm that ensures differential privacy without sacrificing accuracy, and studies its convergence rate and the trade-off with privacy.
https://openreview.net/forum?id=RtnGLJNtEG
Compressor summary: The paper analyzes how local gradients and second-order behavior of deep neural networks affect their robustness against adversarial attacks and proposes a scalable algorithm to compute and use curvature bounds as a regularizer for training.
https://openreview.net/forum?id=RtCmp5F9lN
Compressor summary: Physics-aware proxy model (PAPM) is a new approach that combines partial physics knowledge and flexible adaptation for better generalization and performance in process systems modeling, with less computational cost and fewer parameters than existing methods.
https://openreview.net/forum?id=RsIMGYzBcv
Compressor summary: Key points: - A new algorithm for online resource allocation with non-stationary customer arrivals and unknown click-through rates - It combines ideas from stochastic contextual bandits and online matching - It has low regret when customer arrivals are near-stationary and optimal competitive ratio otherwise - It performs well in numerical experiments Summary: The paper presents a novel algorithm that efficiently allocates resources to customers with changing arrival rates and unknown click-through rates, using insights from two fields and showing good performance in various scenarios.
https://openreview.net/forum?id=Rp8R9C0Sth
Compressor summary: AutoOS is a framework using Large Language Models to optimize Linux kernel configurations for AIoT applications automatically and efficiently.
https://openreview.net/forum?id=RnbobOgbn0
Compressor summary: The paper proposes projection-free online convex optimization algorithms that use linear optimization oracles and achieve low regret and constraints violation under time-varying constraints.
https://openreview.net/forum?id=RlibRvH4B4
Compressor summary: The paper proposes a new algorithm called CIAO for open ad hoc teamwork using graph-based policy learning and cooperative game theory to improve explanations and performance.
https://openreview.net/forum?id=RiQbe8RwCe
Compressor summary: The paper shows that small batch sizes do not help online learning, but SGD noise has computational benefits in this setting, supporting a noisy golden path hypothesis.
https://openreview.net/forum?id=RiM3cl9MdK
Compressor summary: CFG is an effective inference-time technique for language modeling that enhances various tasks and preferences, outperforming some models with less parameters.
https://openreview.net/forum?id=RfsagmV1AG
Compressor summary: The paper analyzes biased policy gradient methods in reinforcement learning and their second-order behavior, including vanilla estimators and actor-critic algorithms.
https://openreview.net/forum?id=RfQT6vJt8b
Compressor summary: Blum & Stangl (2019) prove that fair classifiers with equal opportunity constraints can achieve optimal accuracy even on extremely biased data, and the authors extend this result to various settings and applications.
https://openreview.net/forum?id=RbnojVv4HK
Compressor summary: The paper investigates how dataset biases affect dataset condensation and proposes a sample reweighting method using kernel density estimation to reduce bias amplification, achieving significant improvements on benchmark datasets.
https://openreview.net/forum?id=RbiBKPtuHp
Compressor summary: DCMHA is a more efficient and expressive attention mechanism than MHA that improves Transformer performance in language modeling tasks.
https://openreview.net/forum?id=RZHRnnGcEx
Compressor summary: TURTLE is a method that achieves state-of-the-art unsupervised performance on various tasks by searching for the labeling of a dataset using representation spaces of different foundation models without any supervision or task-specific learning.
https://openreview.net/forum?id=RYmmgedVjR
Compressor summary: The ForgetFilter algorithm filters out unsafe data from large language models based on how easily they forget it, ensuring safety without sacrificing performance.
https://openreview.net/forum?id=RXxTuxPopa
Compressor summary: The paper proposes a new method to classify events with reliable uncertainty estimates, accounting for distributional shifts between train and target data by using nuisance parameter-dependent cutoffs and estimating the ROC across the nuisance parameter space.
https://openreview.net/forum?id=RPMTNGMq0O
Compressor summary: The paper proposes a regularization technique for stochastic first-order methods that stabilizes iterates and provides performance guarantees even when gradient noise is potentially unbounded.
https://openreview.net/forum?id=RLENZ8pNnn
Compressor summary: INSTINCT algorithm optimizes instructions for large language models using neural networks and transformers, achieving better performance than baselines in various tasks.
https://openreview.net/forum?id=RLA4JTckXe
Compressor summary: Reverse GNNs improve node classification on heterophilic graphs by inverting message passing and mitigating over-smoothing.
https://openreview.net/forum?id=RKlmOBFwAh
Compressor summary: The paper proposes a new attack that uses certifications to create more effective adversarial examples and questions the security of certification mechanisms in neural network robustness.
https://openreview.net/forum?id=RIMRKeeVsr
Compressor summary: This study investigates how retrieval helps vision-language models adapt to new tasks, revealing insights on uni-modal and cross-modal retrieval and the importance of logit ensemble.
https://openreview.net/forum?id=RI4GA8amUI
Compressor summary: The paper studies how the test error of learning the readout layer depends on the population covariance of features for large neural networks with structured weights and high-dimensional inputs.
https://openreview.net/forum?id=RFhkcqRmTD
Compressor summary: The text proposes a new objective function for classification tasks based on the shifted log (SL) $f$-divergence, which shows better accuracy than existing methods in various applications.
https://openreview.net/forum?id=RDofzHLuX4
Compressor summary: Key points: - Novel regression adjustment method for distributional treatment effects in randomized experiments - Uses pre-treatment covariates and machine learning techniques to improve precision - Valid and inferrable with well-estimated nuisance components - Simulation and real data analysis show effectiveness Summary: The paper proposes a new regression method that uses covariates and machine learning to estimate distributional treatment effects in randomized experiments more precisely and reliably, based on simulations and real data.
https://openreview.net/forum?id=R8nbccD7kv
Compressor summary: Our method detects outliers in unsupervised learning by leveraging the initial memorization of inliers by under-fitted deep generative models, making it fast and effective for various data types.
https://openreview.net/forum?id=R83VIZtHXA
Compressor summary: The paper proposes a method for online reinforcement learning policy learning that handles distribution discrepancies due to policy or dynamics shifts, called Occupancy-Matching Policy Optimization (OMPO), and shows its effectiveness in various environments.
https://openreview.net/forum?id=R6GT1UDcOW
Compressor summary: The paper shows how combining a target network and over-parameterized linear function approximation can improve bootstrapped value estimation with less strict convergence conditions, even for off-policy data.
https://openreview.net/forum?id=R4Ng8zYaiz
Compressor summary: MMT-Bench is a comprehensive benchmark for evaluating large vision-language models on diverse multimodal tasks requiring visual reasoning and localization, covering 32 core meta-tasks and 162 subtasks.
https://openreview.net/forum?id=R1auM3tLPE
Compressor summary: Bandit Class-specific Conformal Prediction (BCCP) is a method that provides coverage guarantees for class predictions in online learning settings with limited label information, using stochastic gradient descent and unbiased estimation.
https://openreview.net/forum?id=R0SoZvqXyQ
Compressor summary: MuxServe is a system that efficiently serves multiple large language models by colocating them based on popularity and managing resources adaptively.
https://openreview.net/forum?id=QwgSOwynxD
Compressor summary: Key points: - Generative models are important but lack a theory for generalization and uncertainty - The paper proposes a bias-variance-covariance decomposition for kernel scores - This allows deriving variance and entropy for uncertainty estimation using generated samples only - The framework works for image, audio, and language generation Summary: The paper presents a new theoretical framework to estimate uncertainty in generative models using kernel scores and their decomposition, which can be applied to various domains and model types.
https://openreview.net/forum?id=QvABoVGdRp
Compressor summary: This paper proposes a new method to improve the robustness of Spiking Neural Networks against adversarial attacks by regularizing their gradient sparsity, which is theoretically proven and experimentally validated.
https://openreview.net/forum?id=Qv5szC1zp7
Compressor summary: The paper introduces a new online learning algorithm for CMDPs with long-term constraints that can handle stochastic and adversarial rewards and constraints without knowing the transition function, achieving competitive performance in both settings.
https://openreview.net/forum?id=QhqQJqe0Wq
Compressor summary: Score identity Distillation (SiD) is a data-free method that quickly improves the performance of pretrained diffusion models using three identities and a novel loss mechanism, achieving high efficiency and quality in generation.
https://openreview.net/forum?id=QhKsE7YAJk
Compressor summary: The paper presents a polynomial-time algorithm to determine if Naive Bayes Classifiers are certifiably robust to missing values in dirty datasets, and shows that data poisoning attacks are easier for single than multiple test points.
https://openreview.net/forum?id=QhHMx51ir6
Compressor summary: SMAT is a method that improves vision foundation models' transfer abilities by automatically isolating subsets of pre-trained parameters for meta-tuning on each task, overcoming OOD sensitivity and achieving state-of-the-art results.
https://openreview.net/forum?id=QgvBcOsF4B
Compressor summary: Key points: - The paper proposes M$^2$FedSA, a method to train large-scale multimodal models under FL with privacy protection and efficiency improvement. - M$^2$FedSA uses Split Learning, specialized adapters, and modality knowledge transfer to achieve its goals. - The paper evaluates M$^2$FedSA on various multimodal classification tasks and releases the code. Summary: M$^2$FedSA is a novel method for training large-scale multimodal models in a privacy-preserving and efficient way, using Split Learning, adapters, and modality knowledge transfer. It shows promising results on several multimodal classification tasks.
https://openreview.net/forum?id=QgMqvxvWpX
Compressor summary: The text discusses how functional equivalence helps reduce the complexity of neural networks, make them easier to train, and provides insights into overparameterization, generalization, and optimization.
https://openreview.net/forum?id=Qc5umSsUi8
Compressor summary: The paper proposes a novel algorithm for safe policy improvement in multi-agent domains using Monte Carlo Tree Search, factorization, and Max-Plus constraint.
https://openreview.net/forum?id=Qb68Rs0p9f
Compressor summary: The authors propose a new method for learning potential-based motion planning using neural networks that can optimize trajectories, avoid local minima, and handle various constraints.
https://openreview.net/forum?id=QZgo9JZpLq
Compressor summary: Summary: SSMs, a potential alternative for large language models, have similar expressive limitations as transformers in solving state-tracking problems despite their recurrent formulation.
https://openreview.net/forum?id=QZd3rvlP76
Compressor summary: The paper proposes a new self-attention layer for tabular data using matrix polynomials to address the oversmoothing issue in Transformers and improve model scalability.
https://openreview.net/forum?id=QZ1DVzr6N9
Compressor summary: The paper proposes a method to optimize the filter parameter of the Mapper graph in topological data analysis, which improves the representation and visualization of data structures.
https://openreview.net/forum?id=QXqXGDapkQ
Compressor summary: The authors present SleepFM, a multi-modal foundation model for sleep analysis that outperforms standard methods on sleep stage classification and sleep disordered breathing detection using a novel contrastive learning approach.
https://openreview.net/forum?id=QXEx16jWdN
Compressor summary: The paper proposes an improved version of adversarial EBMs by using denoising steps and a variational posterior distribution, leading to better generation and density estimation.
https://openreview.net/forum?id=QTt2xJI8vk
Compressor summary: The paper proposes an adaptive domain partitioning algorithm for reward-free RL that uses kernel-based function approximations and achieves order-optimal sample complexity for a wide range of kernels.
https://openreview.net/forum?id=QRjTDhCIO8
Compressor summary: The paper introduces Re-Dock, a new deep learning method for predicting protein-ligand binding structures that considers both ligand and pocket sidechain conformations, and shows its superior performance on benchmark datasets.
https://openreview.net/forum?id=QRDfBIhrJq
Compressor summary: MFRNP is a new multi-fidelity surrogate modeling framework that improves scalability, inference performance, and accuracy by explicitly modeling the residual between lower and higher fidelity data using neural network encoders, decoders, and optimized lower fidelity decoders.
https://openreview.net/forum?id=QQkK6YH0Th
Compressor summary: The paper proposes a novel online learning scheme that adapts between privacy-preserving recommendations and commitment mechanisms over multiple rounds, achieving low regret in multi-round mechanism design problems.
https://openreview.net/forum?id=QPy7zLfvof
Compressor summary: The paper proposes a deep-learning framework for tabular data that predicts interpretable cluster assignments with feature selection.
https://openreview.net/forum?id=QPsEPI9bvp
Compressor summary: The paper explores minimax optimization algorithms that work under a relaxed Lipschitz smoothness condition called generalized smoothness, which allows them to converge and have better theoretical guarantees in more machine learning applications.
https://openreview.net/forum?id=QMy2RLnxGN
Compressor summary: DoraemonGPT is a system that uses LLMs to understand dynamic scenes, especially videos, and perform tasks across different domains with the help of specialized tools and a novel planner based on Monte Carlo Tree Search.
https://openreview.net/forum?id=QLtxj3erlJ
Compressor summary: The paper proposes a cheap stochastic iterative method for solving problems with generalized orthogonality constraints using random estimates of $B$, achieving similar convergence rates as traditional methods but with lower cost and complexity.
https://openreview.net/forum?id=QLcBzRI3V3
Compressor summary: RiC is a simple and adaptable method for aligning foundation models with human preferences using supervised fine-tuning, which reduces the cost and complexity of previous approaches.
https://openreview.net/forum?id=QLOvxGwbIM
Compressor summary: The paper introduces Bayesian Power Steering (BPS), a novel network structure for fine-tuning large diffusion models that efficiently extracts task-specific knowledge and outperforms contemporary methods in various tasks.
https://openreview.net/forum?id=QKnWXX3aVm
Compressor summary: The paper proposes algorithms to improve generalizing RCT results to a target population by using an additional observational study, while accounting for high-quality, low-quality, and confounded data.
https://openreview.net/forum?id=QJkG8Mln72
Compressor summary: DPZero is a new private zeroth-order algorithm for fine-tuning large language models that reduces memory demands and protects sensitive information.
https://openreview.net/forum?id=QH4mXDEULp
Compressor summary: DiGS is a new sampling method that effectively deals with distant and disconnected modes in multi-modal distributions using Gaussian convolution and Gibbs sampling, improving performance in tasks like Bayesian inference and molecular dynamics.
https://openreview.net/forum?id=QGAeWRRe6e
Compressor summary: The paper proposes a systematic approach to optimize watermark trade-offs for generative LLMs, leading to better robust and efficient watermarks.
https://openreview.net/forum?id=QFMcXz6e4Y
Compressor summary: Flexible meta pruning (FMP) is a lightweight image super-resolution method that simultaneously prunes network channels and weights using hypernetwork-generated meta-data, achieving flexible and structured sparsity control.
https://openreview.net/forum?id=QE6iC9s6vU
Compressor summary: The text discusses using graph neural networks (GNNs) to predict river discharge in a network of gauge stations, but finds that GNNs do not benefit from the river network topology information and struggle with sudden spikes.
https://openreview.net/forum?id=QCZabhKQhB
Compressor summary: Transformers can efficiently simulate and be simulated by communication rounds, enabling them to solve basic computational tasks faster than other neural sequence models.
https://openreview.net/forum?id=QBj7Uurdwf
Compressor summary: The text discusses various methods to learn representations of Recurrent Neural Network (RNN) weights and evaluates their performance on downstream tasks, with functionalist approaches showing better results.
https://openreview.net/forum?id=QAGRPiC3FS
Compressor summary: RigorLLM is a novel framework that efficiently and effectively moderates harmful and unsafe inputs and outputs for Large Language Models using a multi-faceted approach.
https://openreview.net/forum?id=Q8uJyOwOsd
Compressor summary: This paper introduces a new method for scribble-supervised semantic segmentation that uses feature prototypes to improve performance and reduce annotation costs.
https://openreview.net/forum?id=Q3104y8djk
Compressor summary: CogBench is a benchmark for evaluating large language models based on cognitive psychology experiments, revealing insights about their behavior, performance, and alignment with humans.
https://openreview.net/forum?id=PzjDsfYwLC
Compressor summary: The text discusses evaluating vision language models' (VLMs) composition skills using game theory, finding weaknesses in their reasoning abilities and providing guidance for future research.
https://openreview.net/forum?id=PykISfqvet
Compressor summary: The paper presents two density ratio estimation methods with outlier-robustness based on divergences, one convex and one DC, and demonstrates their superior performance in heavy contamination scenarios.
https://openreview.net/forum?id=PxHmxoFOgI
Compressor summary: The paper proposes and analyzes new methods for solving constrained nonconvex nonsmooth optimization problems with stochastic zeroth-order algorithms and novel concepts of approximate stationarity.
https://openreview.net/forum?id=PudBRuNa8r
Compressor summary: The paper proposes a new way to train public models that makes their predictions more fair and equitable for different downstream agents using a novel Equitable Objective and policy gradient algorithm.
https://openreview.net/forum?id=Pte6iiXvpf
Compressor summary: This paper proposes general nonparametric methods for learning causal representations from heterogeneous or nonstationary data and shows how they relate to other assumptions like parametric models or hard interventions.
https://openreview.net/forum?id=PrmxFWI1Fr
Compressor summary: The paper discusses how Bayesian deep learning can enhance deep learning capabilities in various settings and addresses existing challenges and future research directions.
https://openreview.net/forum?id=PpPZ6W7rxy
Compressor summary: Efficient exploration helps improve large language models using human feedback, and double Thompson sampling with epistemic neural networks is a good technique for this purpose.
https://openreview.net/forum?id=PpBs2iL0jv
Compressor summary: The paper proposes a new transfer learning method, ANT, for image generation with limited data, which uses similarity-guided training and adversarial noise selection to improve performance and quality.
https://openreview.net/forum?id=PnyYgWMMwj
Compressor summary: The article explores the finite representation of deep neural networks as composite functions of mappings and proves their universal approximation by a finite vocabulary.
https://openreview.net/forum?id=PlVjIGaFdH
Compressor summary: The paper proposes a new framework for training diffusion models that can handle corrupted data and reduce memorization of the training set.
https://openreview.net/forum?id=PlM30j9i80
Compressor summary: DFR-Det is a method to improve 1-bit detectors' ability to detect tiny objects in aerial images by refining feature representation using an information bottleneck and a new decoder with a foreground mask.
https://openreview.net/forum?id=PjiRSyUt7e
Compressor summary: ConTextual is a new dataset for evaluating multimodal models' ability to reason over text and visual elements in context-rich images, revealing a significant gap between current models and human performance.
https://openreview.net/forum?id=PjVqEErDgK
Compressor summary: Prospector heads are an efficient and interpretable feature attribution method that works across different data modalities and outperforms baseline methods.
https://openreview.net/forum?id=Pbey7LqBRl
Compressor summary: The paper proposes enhancing graph neural networks with smoothed particle hydrodynamics components to improve their performance in simulating fluid dynamics.
https://openreview.net/forum?id=Paw0BkPaTN
Compressor summary: The text introduces a new method for generating universal adversarial perturbations (UAPs) that are robust to real-world transformations, such as rotation and contrast changes, which can improve the effectiveness of practical attacks on deep neural networks.
https://openreview.net/forum?id=Pa3GyTe3kf
Compressor summary: This paper explores whether large language models can help speed up Bayesian optimization in material discovery and finds that they can, but only if they are trained on relevant domain data.
https://openreview.net/forum?id=PYDCwWvbG7
Compressor summary: The paper introduces a new Quantum-based Matching Kernel (QBMK) for un-attributed graphs, which captures both global and local structural characteristics and outperforms existing methods in experiments.
https://openreview.net/forum?id=PY3bKuorBI
Compressor summary: This paper proposes a unified theory to bound the excess risk of kernel regression for realistic settings and shows that kernel methods have a built-in self-regularization mechanism.
https://openreview.net/forum?id=PTGJOUlQ68
Compressor summary: The paper proposes an optimal private vector mean estimation protocol in the shuffle model of privacy, and studies its properties in various settings.
https://openreview.net/forum?id=PSzyBN7LIA
Compressor summary: The paper proposes a novel algorithm, ME-DOL, that achieves the optimal convergence rate for decentralized nonsmooth nonconvex stochastic optimization in finite time.
https://openreview.net/forum?id=PSQ5Z920M8
Compressor summary: The paper proposes a working memory module for transformer-based agents to improve their efficiency and generalization by mitigating the forgetting phenomenon and blending information from different tasks.
https://openreview.net/forum?id=PQWVUbqQtQ
Compressor summary: The paper suggests using the reasonable person standard from law to guide the development and evaluation of AI behavior that emulates human norms.
https://openreview.net/forum?id=PQ0ERKKYJu
Compressor summary: Satellite data is a unique modality for machine learning that requires a new research agenda to address its challenges and opportunities.
https://openreview.net/forum?id=PPoQz8K4GZ
Compressor summary: PVA is a framework that uses semantic information from text to align images across domains, enabling zero-shot policy transfer in RL with limited cross-domain data.
https://openreview.net/forum?id=PNsdnl8blk
Compressor summary: MAgg is a machine learning method that uses metamorphic relations to improve prediction aggregation for combinatorial problems.
https://openreview.net/forum?id=PMASooqgoq
Compressor summary: The paper proposes a new density estimation method that regularizes a Sobolev norm, is statistically consistent and interpretable, and performs well on an anomaly detection benchmark.
https://openreview.net/forum?id=PLAGGbssT8
Compressor summary: Retro 48B is a large language model pretrained with retrieval that outperforms GPT 43B on various zero-shot tasks and can improve instruction tuning.
https://openreview.net/forum?id=PKdege0U6Z
Compressor summary: GeoMix uses Gromov-Wasserstein distance to generate consistent synthetic samples for graph data, improving GNN performance.
https://openreview.net/forum?id=PKJqsZD5nQ
Compressor summary: RICE is a new method for improving deep reinforcement learning agents' performance by using explanation techniques to create a better initial state distribution, leading to better exploration and sub-optimality bounds.
https://openreview.net/forum?id=PHjkVjR78A
Compressor summary: The paper proposes a method to teach large multi-modality models (LMMs) with text-defined rating levels instead of scores, achieving state-of-the-art accuracy on image and video quality assessment tasks.
https://openreview.net/forum?id=PHUAG63Efe
Compressor summary: AegisFL is a privacy-preserving federated learning system that enables flexible robust aggregation algorithms while maintaining efficiency and preventing model exposure.
https://openreview.net/forum?id=PG5fV50maR
Compressor summary: LESS is an algorithm that selects relevant instruction data for large language models to develop specialized skills like reasoning by estimating data influences and searching for low-rank gradient similarity.
https://openreview.net/forum?id=PEpbUobfJv
Compressor summary: Medusa is a method that accelerates language model inference by predicting multiple tokens in parallel using extra decoding heads and tree-based attention, achieving significant speedup without sacrificing quality.
https://openreview.net/forum?id=PDUQRBPkks
Compressor summary: The paper proposes a distributed algorithm for quantile regression that is robust, efficient, and achieves high accuracy with low communication requirements.
https://openreview.net/forum?id=PDO2Oc1cS1
Compressor summary: HOCTC is a novel network that extends contrastive learning to sparse ordinal tensor data by using attention-based query-expansion and fine-grained comparisons for high-order representation learning.
https://openreview.net/forum?id=PApqOVbHYF
Compressor summary: The paper presents bootstrap inference methods for LFM and FPPE models and develops a new procedure for bootstrapping constrained M-estimators using epi-convergence theory, which is tested on synthetic and real data.
https://openreview.net/forum?id=PAbkWU0KDG
Compressor summary: PAIL is a novel imitation learning algorithm that uses limited human preferences to improve policy learning from imperfect demonstrations, achieving better results in sequential decision-making tasks.
https://openreview.net/forum?id=PAPY0cAB3C
Compressor summary: LEAP is a method to improve LLMs' performance on downstream tasks by learning from mistakes made during few-shot learning and applying learned principles to unseen problems.
https://openreview.net/forum?id=P7qwBmzwwZ
Compressor summary: IRM-TV-$\ell_1$ is a novel framework that generalizes invariant features using a total variation model based on the $L^2$ norm, expanding function classes and improving robustness in denoising and out-of-distribution scenarios.
https://openreview.net/forum?id=OrVl8R13Wy
Compressor summary: Sparse Cocktail is a novel sparse co-training framework that trains multiple subnetworks with diverse sparsity patterns and ratios, allowing flexible switching between them based on resource availability during inference.
https://openreview.net/forum?id=OnkA4zaEU9
Compressor summary: The paper proposes a triadic-online change detection framework with certifiable robustness, provable optimality, guaranteed convergence, and asynchronous distributed implementation for various applications.
https://openreview.net/forum?id=OnidGtOhg3
Compressor summary: The Diffusion Model-Augmented Behavioral Cloning (DBC) is a framework that improves imitation learning by modeling both conditional and joint probabilities of expert behaviors, leading to better generalization and performance in continuous control tasks.
https://openreview.net/forum?id=OndZHBUA1G
Compressor summary: The paper analyzes theoretical guarantees of classical optimization methods for problems with biased gradient oracles, showing their invariance to errors and applicability to different settings.
https://openreview.net/forum?id=OnOaj3g9fi
Compressor summary: The proposed framework adapts compute allocation for score estimation in diffusion models, improving sampling speed without sacrificing quality.
https://openreview.net/forum?id=OnEaBGU3LO
Compressor summary: The paper proposes FRAG, a method to improve video quality by preserving high-frequency components during denoising, without requiring extra training.
https://openreview.net/forum?id=Olix9pk6nV
Compressor summary: The paper introduces a novel linear programming framework for reward learning from human demonstrations and feedback, with provable sample efficiency and optimality guarantees.
https://openreview.net/forum?id=OkChMnjF6s
Compressor summary: The paper explores the fragility of machine unlearning verification strategies and proposes two novel adversarial methods to bypass them, revealing the need for more research on the safety of machine unlearning.
https://openreview.net/forum?id=OjBW993g79
Compressor summary: MGit is a system that helps manage related machine learning models by recording their relationships and optimizing storage and testing processes, reducing storage size and update time.
https://openreview.net/forum?id=Oj18qGN1gC
Compressor summary: The paper proposes a new algorithm, SEA, that improves sparse support recovery using the straight-through estimator and analyzes its performance under the Restricted Isometry Property.
https://openreview.net/forum?id=OiI12sNbgD
Compressor summary: The paper introduces Atari Pre-training Benchmark (Atari-PB), a unified benchmark to evaluate pre-training methods in vision-based Reinforcement Learning, and shows that objectives focused on task-agnostic features improve generalization across diverse environments.
https://openreview.net/forum?id=OgG0I5toZZ
Compressor summary: The paper proposes asking users questions about both features and comparisons when learning their reward functions from preference data, leading to faster and more accurate rewards in different domains.
https://openreview.net/forum?id=OfT8MgIqHT
Compressor summary: The paper proposes a simple scaling of the Gaussian process lengthscale prior to improve Bayesian optimization performance in high-dimensional problems, and shows that standard BO works better than previously thought in this setting.
https://openreview.net/forum?id=OenMwDPqWn
Compressor summary: The paper proposes a method to generate more descriptive and accurate explanations for model predictions using semantic graphs and graph neural networks, outperforming previous approaches in both quantitative and qualitative evaluation.
https://openreview.net/forum?id=OdsZS0E0AO
Compressor summary: The paper introduces a new way to check how good optimization proxies are, which works faster and for more types of problems.
https://openreview.net/forum?id=OdPlFWExX1
Compressor summary: The paper introduces a new family of sparse Hopfield networks with end-to-end differentiable transformations and shows their applications in pattern retrieval tasks.
https://openreview.net/forum?id=OYw6sS8QmL
Compressor summary: The paper proposes a novel Bayesian model-free method, the Bayesian exploration network (BEN), which can learn true Bayes-optimal policies by modelling both types of uncertainty and achieving optimal trade-offs between them.
https://openreview.net/forum?id=OYL91MHfuU
Compressor summary: The paper introduces Controllable Prompt Tuning (CPT), a method that optimizes performance across different groups or domains without sacrificing performance on any of them, using prompt-tuning techniques and minimal tunable parameters.
https://openreview.net/forum?id=OXzkw7vFIO
Compressor summary: The paper proposes a method for counterfactual image editing using causal language and neural causal models, while acknowledging the limitations of the task.
https://openreview.net/forum?id=OVn8FpeBpG
Compressor summary: The paper shows wide and deep ReLU neural network classifiers are consistent and optimal for various function classes, including those without smoothness assumptions.
https://openreview.net/forum?id=OURP5Z58jt
Compressor summary: Strategic classification studies how to learn decision rules that are robust to input manipulation, and this paper focuses on one-shot settings where there's uncertainty in users' costs and aims to design efficient algorithms that minimize worst-case risk.
https://openreview.net/forum?id=OTmcsyEO5G
Compressor summary: ReadAgent is a system that enhances large language models' ability to read long documents by using memory episodes and gist memories, leading to improved performance on reading comprehension tasks with extended context windows.
https://openreview.net/forum?id=OS5dqxmmtl
Compressor summary: SparQ Attention is a technique that improves the speed and efficiency of large language models without changing their pre-training or fine-tuning by using memory bandwidth more effectively in attention layers.
https://openreview.net/forum?id=OS0szhkPmF
Compressor summary: The paper proposes a self-supervised model that learns disentangled graph representations to improve out-of-distribution generalization for graph neural networks without task-dependent labels.
https://openreview.net/forum?id=OQ97v7uRGc
Compressor summary: The paper studies finding stationary points in bilevel optimization when the lower-level problem is unconstrained and strongly convex, proposing a first-order method that converges using $y^*$-aware oracles and showing upper and lower bounds for its complexity.
https://openreview.net/forum?id=OQ7TlOphGX
Compressor summary: SSWDP is a novel self-supervised method that predicts distortion in observed trajectories to improve spatio-temporal representation learning for trajectory prediction.
https://openreview.net/forum?id=ONOtpXLqqw
Compressor summary: LongRoPE extends LLMs' context window to 2048k tokens using efficient search, progressive extension, and readjustment, while maintaining performance on short context windows.
https://openreview.net/forum?id=OMKNBzf6HJ
Compressor summary: LFIS is a flow-based model that uses neural networks to learn velocity fields and generate unbiased samples from complex density functions.
https://openreview.net/forum?id=OLvgrLtv6J
Compressor summary: The paper proposes SymC, a code semantics model for Large Language Models that leverages code symmetries for efficient and accurate program analysis, outperforming GPT-4 without pre-training.
https://openreview.net/forum?id=OKYfaYQlML
Compressor summary: Key points: - VFMs are powerful but costly to run - The paper proposes a knowledge transfer approach to train small task-specific models from VFMs with limited data - The approach outperforms other methods and reduces compute cost - A retrieval-augmented strategy is introduced to curate effective transfer sets Summary: The paper presents a knowledge transfer method that trains small task-specific models from large VFMs with limited data, achieving better performance and lower compute cost than existing approaches, and using web-scale image retrieval to select transfer sets.
https://openreview.net/forum?id=OJTKlubFk1
Compressor summary: The paper proposes a novel error-feedback technique that compresses gradient information for full-matrix preconditioning in deep learning, reducing storage costs significantly without sacrificing performance.
https://openreview.net/forum?id=OI1YP53WKI
Compressor summary: ReDiffuser improves diffusion models by using confidence estimation from Random Network Distillation for reliable decision-making in offline reinforcement learning.
https://openreview.net/forum?id=OHFxcU9jwW
Compressor summary: The paper proposes a diffusion-based network to generate missing views for incomplete multi-view data and a data augmentation strategy to improve clustering performance.
https://openreview.net/forum?id=OF7e0w1uon
Compressor summary: HyperAgent is a reinforcement learning algorithm that uses the hypermodel framework for efficient exploration, achieving robust performance in large-scale deep RL benchmarks and low per-step computational complexity.
https://openreview.net/forum?id=OERwuPzHdh
Compressor summary: The paper proposes a new method called DNA-SE that uses deep neural networks to solve semiparametric estimation problems involving Fredholm integral equations, improving both numerical and statistical performance over traditional methods.
https://openreview.net/forum?id=OBs0AjXE3F
Compressor summary: The paper proposes KV-Runahead, a parallelization scheme to speed up the first token generation in large language models by using a key-value cache that is already used for token generation.
https://openreview.net/forum?id=O8rrXl71D5
Compressor summary: The text explains how a novel causal framework helps understand the diversity, emergence dynamics, and subcircuits involved in induction heads, which are critical for in-context learning in transformer models.
https://openreview.net/forum?id=O6tenHWTUU
Compressor summary: The paper proposes a representation-based framework for tackling partially observable reinforcement learning problems that improves efficiency and performance.
https://openreview.net/forum?id=O4nXWHPl6g
Compressor summary: The paper proposes a new approach, Class-Conditional Knowledge Distillation (CCKD), and its variant AACCKD, to improve generalization on class-imbalanced data by learning the teacher model's class-conditional probability distribution and enhancing feature learning.
https://openreview.net/forum?id=O4cHTxW9BS
Compressor summary: SPIN is a new fine-tuning method that improves Large Language Models by letting them play against themselves and generate their own training data, achieving better results than other methods even without extra human-annotated data.
https://openreview.net/forum?id=O45u81aby2
Compressor summary: ToRES is an IMVC method that uses prototype-sample affinity, view-wise and cross-view prototypes, and unified representation learning and clustering to overcome common drawbacks of existing IMVC methods.
https://openreview.net/forum?id=O3CFN1VIwt
Compressor summary: The paper explores task correlations in graph self-supervised learning, proposes Graph Task Correlation Modeling (GraphTCM) to improve representation quality, and shows its effectiveness on downstream tasks.
https://openreview.net/forum?id=O1hmwi51pp
Compressor summary: The paper presents an automated loss function search framework for class-imbalanced node classification tasks on graphs, which improves performance and generalizes well across different datasets and network structures.
https://openreview.net/forum?id=Nxz3CDtGXp
Compressor summary: The paper evaluates how well balancing strategies work for counterfactual estimation with time series data, finding that their effectiveness may need reconsideration.
https://openreview.net/forum?id=NwYsuFuelg
Compressor summary: DynaBRO is a dynamic fault-tolerant machine learning method that can handle changing Byzantine behaviors and achieve near-optimal convergence rates using multi-level Monte Carlo gradient estimation and adaptive learning rate.
https://openreview.net/forum?id=Nw7yOe8nBi
Compressor summary: The authors train a differentially private image captioner on a large dataset and show that it can learn high-quality image features for various downstream tasks, challenging the belief that such privacy-preserving representation learning is not possible.
https://openreview.net/forum?id=NvBJOcmti6
Compressor summary: The paper introduces graded non-convexity, a concept that divides non-convex problems into subclasses, and proposes gradient methods with spectral preconditioning that improve convergence rates for these problems.
https://openreview.net/forum?id=Nue7KgVZ6e
Compressor summary: The text proposes multigroup robust algorithms that provide more meaningful robustness guarantees for different subpopulations by accounting for how data corruption affects each group differently.
https://openreview.net/forum?id=NsHxeSCtgr
Compressor summary: This paper proposes a framework, LIDAO, that debias large language models while maintaining better fluency and robustness against adversarial prompts.
https://openreview.net/forum?id=Nm6jYZsBum
Compressor summary: The paper introduces Multimodal Composition Learning, a method to improve frozen Large Language Models' performance in multimodal tasks by using two specialized tasks: MC-Cap and MC-Ret.
https://openreview.net/forum?id=NlM4gp8hyO
Compressor summary: Chunked-TD is a model-based reinforcement learning method that uses world models to adjust the bias-variance tradeoff in TD($\lambda$) and speed up credit assignment.
https://openreview.net/forum?id=Nl3RG5XWAt
Compressor summary: Topological deep learning (TDL) is a new area of machine learning that uses topological features to design and understand deep models, with many open problems and opportunities for research.
https://openreview.net/forum?id=NkN6wrYXe5
Compressor summary: The paper proposes a novel federated multi-level compositional minimax algorithm to improve AUC maximization in imbalanced data classification problems with rigorous theoretical guarantees and shows its effectiveness through empirical evaluations.
https://openreview.net/forum?id=NgaYcefBnZ
Compressor summary: Key points: - Machine learning algorithms face distributional shifts and over-pessimism issues. - The paper proposes Geometry-Calibrated DRO (GCDRO) for regression, which incorporates data geometry into calibration terms in DRO. - The paper connects the risk objective to the Helmholtz free energy and develops an approximate minimax optimization algorithm. - Experiments show GCDRO outperforms conventional DRO methods. Summary: The paper introduces Geometry-Calibrated DRO, a novel method for regression that mitigates over-pessimism in distributionally robust optimization by using data geometry and free energy principles.
https://openreview.net/forum?id=NeotatlYOL
Compressor summary: The paper proposes a new method, SpikeZIP-TF, that converts ANNs to SNNs with no accuracy loss and achieves better results than existing Transformer-based SNNs on CV and NLP tasks.
https://openreview.net/forum?id=NeO2hoSexj
Compressor summary: The paper introduces ALH, an algorithm that combines reinforcement learning with weak environment descriptions (hypotheses) to improve performance and reduce bias in value-based RL.
https://openreview.net/forum?id=NeEbsvnaWE
Compressor summary: This article studies how to estimate privately the probability density of smooth functions in high dimensions and proposes a data-driven approach to choose the best estimator.
https://openreview.net/forum?id=NbYAmsFJrc
Compressor summary: The Trajectory Aggregation Tree (TAT) improves the reliability and stability of diffusion planners by aggregating information from historical and current trajectories in a dynamic tree-like structure.
https://openreview.net/forum?id=NbOlmrB59Z
Compressor summary: The SimPro framework is a novel semi-supervised learning approach that adapts to unknown class distributions in unlabeled data without predefined assumptions, using a probabilistic model and separating conditional and marginal class distributions.
https://openreview.net/forum?id=NZgbwzaOIx
Compressor summary: The paper proposes a framework to understand and improve knowledge distillation by using rank-based loss instead of KL divergence, which is sensitive to model calibration.
https://openreview.net/forum?id=NZQkumsNlf
Compressor summary: NExT-GPT is an end-to-end system that enables any-to-any multimodal understanding and generation by connecting a large language model with adaptors and diffusion decoders, requiring minimal additional training and using modality-switching instructions to facilitate cross-modal semantic understanding.
https://openreview.net/forum?id=NV0q2jdwo0
Compressor summary: The paper proposes an improved feedforward network module for vision transformers that reduces computational cost by enhancing non-linearity using the AGeLU function and a spatial enhancement part, achieving comparable accuracy with fewer parameters and floating-point operations.
https://openreview.net/forum?id=NUlyqMyhO9
Compressor summary: The paper proposes LoCoCo, a method that compresses long context sequences in LLMs by adaptively blending previous and incoming tokens using convolutional kernels, improving efficiency and accuracy.
https://openreview.net/forum?id=NUAbSFqyqb
Compressor summary: The paper presents DPLM, a protein language model that generates diverse and plausible protein sequences using diffusion probabilistic pre-training and can be fine-tuned for various tasks or conditioned on different inputs.
https://openreview.net/forum?id=NS8z5FinYl
Compressor summary: The paper proposes a Bayesian planning method that uses uncertainty estimates from neural networks to improve online planning, and shows its effectiveness on maze and leaper environments.
https://openreview.net/forum?id=NQn2tYLv5I
Compressor summary: The paper introduces new tools to analyze meta-learning on sequences and shows how error decays in different settings.
https://openreview.net/forum?id=NQ6KDfSDFK
Compressor summary: The study proposes two variants of MoLA, a method to mitigate conflicts in heterogeneous data training for unified models aiming at artificial general intelligence.
https://openreview.net/forum?id=NKirMgDsut
Compressor summary: A small modification to stochastic gradient descent with momentum (SGDM) improves its performance in training neural networks with irregular loss functions by scaling the update with an exponentially distributed random scalar.
https://openreview.net/forum?id=NFEJQn7vX0
Compressor summary: The paper studies how to use public data to improve machine learning models while preserving privacy, and develops new algorithms that achieve better performance than existing ones.
https://openreview.net/forum?id=NEv8YqBROO
Compressor summary: The paper proposes LoRA+, a corrected version of LoRA that improves finetuning speed and performance by setting different learning rates for adapter matrices in large width networks.
https://openreview.net/forum?id=NCjlFw1Ab0
Compressor summary: The study proposes a new model for neural working memory using traveling wave dynamics in Recurrent Neural Networks (RNNs), which improves data storage and learning by mimicking the brain's information processing.
https://openreview.net/forum?id=NCT3w7VKjo
Compressor summary: GPL is a new method for positive-unlabeled learning on graph data that reduces edge heterophily to improve classifier training.
https://openreview.net/forum?id=NBAc36V00H
Compressor summary: QINCo is a neural vector quantization method that builds custom codebooks per step, leading to improved data compression and search accuracy compared to conventional methods.
https://openreview.net/forum?id=N6A6t6xlKm
Compressor summary: The paper proposes using convergence rate as a proxy for uncertainty in iterative network architectures, which can improve accuracy and reduce computational cost compared to other methods.
https://openreview.net/forum?id=N3ZrpSCJcJ
Compressor summary: The authors propose a method to improve generative machine learning for molecular systems by using coarse-grained simulations with active learning and conditioning normalizing flows on the coarse-grained space, achieving significant speedup compared to existing approaches.
https://openreview.net/forum?id=N1BPyf7wC2
Compressor summary: The paper studies a model where online agents join dynamically and offline agents value their coverage over them, with limited capacities; it proposes two matching policies and analyzes their competitive ratios depending on capacity and coverage bounds.
https://openreview.net/forum?id=N0ntTjTfHb
Compressor summary: The paper proposes MACURA, a model-based reinforcement learning algorithm that adapts rollout lengths based on local model uncertainty, improving data efficiency and performance.
https://openreview.net/forum?id=Mz1lcJPymz
Compressor summary: The text discusses how binary betting markets aim to balance profit and information, introduces online learning methods for price-setting, and analyses the tradeoff between these goals.
https://openreview.net/forum?id=MwQ53xAIPs
Compressor summary: FasterCUCB is a sublinear time algorithm for matroid semi-bandits that maximizes expected cumulative linear rewards with low regret.
https://openreview.net/forum?id=Mw8kNVfdMs
Compressor summary: The authors propose new evaluation protocols to measure how well generative models capture attributes of images in the training dataset, revealing strengths and weaknesses of existing models.
https://openreview.net/forum?id=Mv8y13wfDm
Compressor summary: The text proposes a method to measure and compare socio-technical risks of foundation models using statistical testing based on stochastic dominance and risk-aware model selection.
https://openreview.net/forum?id=MurkwIl0h3
Compressor summary: BOSS is a method for selecting a balanced subset of diverse and difficult data samples for efficient deep learning model training, considering the subset size and using a novel Beta-scoring importance function.
https://openreview.net/forum?id=MsnJl6JkZS
Compressor summary: The paper proposes ELA, a method to prevent concept bleeding in image generation models by localizing and anchoring entities in their expected regions using auxiliary networks.
https://openreview.net/forum?id=Msjovr9hUe
Compressor summary: The paper proposes SUWR, a local feature selection method that avoids misleading explanations by ensuring no label or feature leakage in complex models.
https://openreview.net/forum?id=MrNq6rbcUi
Compressor summary: The text discusses how to create provably robust prediction sets using conformal prediction that resist adversarial examples and perturbed calibration data.
https://openreview.net/forum?id=MoTUdh9ZCc
Compressor summary: DeCoOp is a novel prompt tuning approach for vision-language models that uses new-class detectors and sub-classifiers to improve performance on base and new classes in open-world settings.
https://openreview.net/forum?id=MmZJ3kJXjX
Compressor summary: The paper proposes a method to evaluate pretrained model representations without real-world data by using synthetic binary classification tasks with Gaussian mixtures, which correlates with actual performance on downstream tasks and helps balance robustness and accuracy.
https://openreview.net/forum?id=MlzUD5CKvZ
Compressor summary: The paper proposes a post-processing framework called R3, which improves the interpretability and accuracy of ProtoPNet, a method for image classification that uses meaningful parts of images.
https://openreview.net/forum?id=MjGCD8wk1k
Compressor summary: LaMAGIC is a language model-based topology generation model that efficiently designs optimized analog circuits from custom specifications using supervised finetuning.
https://openreview.net/forum?id=MikandLqtW
Compressor summary: KnowRLM is a novel Machine Learning-assisted Directed Evolution method that uses a Knowledge Graph of biochemical relationships among amino acids to guide the search for optimal protein variants.
https://openreview.net/forum?id=MgTzMaYHvG
Compressor summary: SafeCoder is a tool that improves the security of code generated by language models without sacrificing their usefulness, by fine-tuning them with a large dataset of secure code.
https://openreview.net/forum?id=Mfk6ZbD6eY
Compressor summary: Stereo Risk is a new deep-learning method for stereo matching that uses continuous risk minimization to estimate scene depth more accurately than existing methods.
https://openreview.net/forum?id=MdPBVWTfwG
Compressor summary: FlashAttention reduces Transformer's attention complexity by being I/O-aware, and this paper proves its optimality for certain memory hierarchies and introduces a new communication complexity protocol for matrix compression.
https://openreview.net/forum?id=MZkqjV4FRT
Compressor summary: Key points: - MAG is a graphical model for causal relations with latent variables - Only one identifiable MEC of MAGs from observational data - No efficient methods for MAG listing except brute force - Propose a new method that avoids brute force and lists all MAGs in MEC - Method is based on recursive determination of local structures of vertices Summary: The paper proposes a novel method to list all the causal graphs in a class, without using brute force, by recursively finding valid local transformations of vertices.
https://openreview.net/forum?id=MWZWUyfFHC
Compressor summary: TinyTrain is an on-device training approach that reduces training time by selectively updating parts of the model and handling data scarcity, achieving high accuracy with lower computation and memory costs.
https://openreview.net/forum?id=MWTicAxmRP
Compressor summary: Our proposed framework for multi-agent policy gradient methods uses clipping to enable optimistic updates that prevent overgeneralization and improve performance on various cooperative learning tasks.
https://openreview.net/forum?id=MV2b44zDd3
Compressor summary: The paper proposes an effective estimator for adversarial risk in non-parametric settings with arbitrary norms and mild regularity conditions, achieving a minimax excess risk of O(sqrt{d/n}) for linear classifiers.
https://openreview.net/forum?id=MUXTt9Yr4T
Compressor summary: The text introduces PromptGIP, a universal model for image processing that uses visual prompting question answering to handle various tasks without task-specific finetuning.
https://openreview.net/forum?id=MSMKQuZhD5
Compressor summary: Amortized varitional deep kernel learning (AVDKL) improves over standard methods for tabular data, node classification, and image recognition by preventing spurious correlations in training.
https://openreview.net/forum?id=MSFxOMM0gK
Compressor summary: The paper presents a faster algorithm for approximating the Balanced Cut problem in a semi-random graph model with adversarial edge modifications.
https://openreview.net/forum?id=MRYS3Zb4iV
Compressor summary: The paper proposes CSIQA, a novel BIQA method that combines global and local perspectives using contrastive learning and attention modules, achieving better performance than existing methods.
https://openreview.net/forum?id=MQirNNU2pC
Compressor summary: The study explores how weight decay affects neuron updates in deep neural networks and its relationship with optimizers, normalization, and learning rate warmup.
https://openreview.net/forum?id=MOrvoYrlOg
Compressor summary: The authors propose an infinite-dimensional approach to optimize machine learning problems and suggest discretizing the algorithm after choosing it, which could lead to new optimization methods for scientific machine learning.
https://openreview.net/forum?id=MMMHufVc2v
Compressor summary: F-design is a non-linear extension of G-optimal design that improves data collection and exploration in various machine learning tasks.
https://openreview.net/forum?id=MKzgqtRtGY
Compressor summary: The authors study how people use text-to-image models in an online game called ArtWhisperer, where users find prompts to generate similar images to a target image, and analyze human-AI interactions, prompt diversity, and AI steerability.
https://openreview.net/forum?id=MKGrRVODWR
Compressor summary: The paper proposes methods to identify causal variables from partially observed data without assuming fixed subsets of latent causes or paired observations, and shows their effectiveness in simulated and real datasets.
https://openreview.net/forum?id=MIRQ3L8vtn
Compressor summary: The paper studies how to detect community structures in private networks using stochastic block models and derives conditions for exact recoverability under edge differential privacy.
https://openreview.net/forum?id=MGkeWJxQVl
Compressor summary: The text proposes RAFA, a framework that combines reasoning and acting with provable regret guarantees using a prompt template and memory buffer to learn and plan in Bayesian adaptive MDPs.
https://openreview.net/forum?id=MFPYCvWsNR
Compressor summary: The paper applies information theory to study generative document retrieval, where documents are indexed by terms and queries are mapped to terms, and proposes a new indexing method that minimizes the bottleneck and improves performance.
https://openreview.net/forum?id=MEZydkOr3l
Compressor summary: The paper proposes MOOSF, a multi-objective optimization method that fuses heterogeneous strategies to improve performance on long-tailed data and resolves potential conflicts between head and tail classes.
https://openreview.net/forum?id=MDAg5Q7IsI
Compressor summary: The text describes a neural model that estimates the entire dose-response curve using the interaction between drug molecules and the tissue transcriptome, outperforming existing models in interpolating and extrapolating inhibitory effects of untried concentrations.
https://openreview.net/forum?id=M8UbECx485
Compressor summary: The text shows how multitask pretraining helps nonlinear neural networks learn useful features by inducing a pseudo-contrastive loss and simplifies binary classification tasks with high dimensions.
https://openreview.net/forum?id=M5ne8enLcr
Compressor summary: The paper proposes HEAL, a framework that uses hypergraphs and line graphs to capture higher-order dependencies for semi-supervised graph classification, outperforming existing methods.
https://openreview.net/forum?id=M5kn9NKIs4
Compressor summary: SemiRES is a semi-supervised framework that combines labeled and unlabeled data for RES, using SAM to improve pseudo-label accuracy and offering different matching strategies for enhanced precision.
https://openreview.net/forum?id=M4ejBhNNrn
Compressor summary: The paper analyzes how Principal Component Regression (PCR) performs on high-dimensional data with realistic assumptions, showing its benefits for generalization and distribution shift.
https://openreview.net/forum?id=M4Htd52HMH
Compressor summary: DeDer is a framework that compresses large language models into small ones for efficient decision-making in embodied tasks using a reasoning-policy and a planning-policy guided by rationales from an embodied knowledge graph.
https://openreview.net/forum?id=M407RM0z6h
Compressor summary: The paper proposes iterated regularization to improve error convergence rates in kernel methods for density ratio estimation, and demonstrates its effectiveness on benchmarks and large-scale evaluations.
https://openreview.net/forum?id=M3uv4qDKOL
Compressor summary: DUPLEX is a new framework for embedding directed graphs that better captures edge information, handles nodes with different connectivity, and adapts to various tasks.
https://openreview.net/forum?id=M3qRRkOuTN
Compressor summary: STEER is a novel method for coordinating asynchronous actions in Multi-Agent Systems, combining hierarchical decision structures, autoregressive sequence models, and exploratory learning techniques.
https://openreview.net/forum?id=M2cwkGleRL
Compressor summary: The authors provide a definition of Large Language Models (LLMs), question some assumptions about them, and suggest areas for further investigation.
https://openreview.net/forum?id=M1ADedSnlJ
Compressor summary: The paper analyzes how guidance affects diffusion models' performance and diversity in generating samples from Gaussian mixture models using theory from differential equations and the Fokker-Plank equation.
https://openreview.net/forum?id=LyJ85kgHFe
Compressor summary: The paper introduces $exttt{MoE-RBench}$, a tool to assess the reliability of Mixture-of-Experts (MoE) models in various dimensions and suggests ways to improve their performance.
https://openreview.net/forum?id=Lwm6TiUP4X
Compressor summary: Lightning Attention is a fast and memory-efficient linear attention implementation that uses different calculation strategies for intra-blocks and inter-blocks, and introduces TransNormerLLM, a new language model architecture tailored to it.
https://openreview.net/forum?id=LwOfVWgEzS
Compressor summary: Key points: - The paper proposes Machine Vision Therapy and Denoising In-Context Learning (DICL) to improve the visual robustness of Multi-modal Large Language Models (MLLMs) under Out-of-Distribution (OOD) scenarios. - Machine Vision Therapy supervises vision models using MLLM predictions, while DICL uses a transition matrix to construct instructions for MLLMs to detect and correct erroneous vision model outputs. - The paper provides theoretical guarantees and experimental results on various OOD datasets. Summary: The paper introduces two methods, Machine Vision Therapy and Denoising In-Context Learning, to enhance the visual robustness of MLLMs by improving their vision models using MLLM predictions and instructions based on a transition matrix. The paper shows theoretical and empirical evidence of their effectiveness under OOD scenarios.
https://openreview.net/forum?id=LvuuYqU0BW
Compressor summary: CDTD is a method for few-shot action recognition that uses causal representation learning to transfer temporally invariant knowledge from pre-trained models and adapts to novel data with limited samples.
https://openreview.net/forum?id=LuhWZ2oJ5L
Compressor summary: S$\Omega$I is a novel method to compute O-information, which measures synergy-redundancy balance in complex multivariate systems, without restrictive assumptions and using a unique model.
https://openreview.net/forum?id=Lt8Lk7IQ5b
Compressor summary: COPAL is an algorithm for pruning large language generative models to adapt them to new domains efficiently and effectively.
https://openreview.net/forum?id=LpAzlcGzJ6
Compressor summary: The paper proposes DFA-RAG, a framework that enhances conversational agents using large language models and a semantic router based on dialogue examples.
https://openreview.net/forum?id=Ln3moCobjO
Compressor summary: The paper proposes novel doubly robust estimators to address sampling selection bias in recommender systems using pseudo-labelings and propensity reconstruction learning with an attention mechanism.
https://openreview.net/forum?id=LmzsgSDkWs
Compressor summary: Key points: - Paper addresses weakly supervised learning with inexact supervision, such as partial labels and unlabeled data - Proposes a novel mutual information-based approach to handle label redundancy and insufficiency - Experiments show superior performance over existing methods in semi-supervised partial label learning and partial-complementary label learning scenarios Summary: The paper presents a new method for weakly supervised learning that handles different forms of inexact supervision by using mutual information to dynamically exchange and filter labels, outperforming existing approaches.
https://openreview.net/forum?id=LlqphyBdeT
Compressor summary: Key points: - Large Language Models (LLMs) are good at natural language but not for specialized domains like physical and biomedical sciences - The authors propose a framework to learn custom input tags that condition the LLM for specialized domains - The input tags consist of domain tags and function tags that help delimit representations and compress instructions - The method enables zero-shot generalization and outperforms expert models in various tasks Summary: The authors present a model-agnostic framework to learn input tags that enhance LLMs for specialized domains by disentangling domains from functions and enabling zero-shot generalization.
https://openreview.net/forum?id=LkJ6qOMv77
Compressor summary: Collage uses multi-component float representation to improve low-precision training by compensating errors and reducing the need for high-precision floating points, achieving faster speedup and less memory usage.
https://openreview.net/forum?id=Ljhrv1Wmbr
Compressor summary: This paper investigates why deep neural networks struggle with out-of-distribution generalization and proposes a new mechanism, feature contamination, that differs from spurious correlations.
https://openreview.net/forum?id=Lhb39btw16
Compressor summary: The paper proposes F2E, a new loss function for protein folding models that improves accuracy in modeling antibody-antigen complexes by optimizing rotational and translational errors between frames.
https://openreview.net/forum?id=LhNsSaAKub
Compressor summary: The paper proposes an unsupervised method to pre-train generalist policies that can adapt quickly to various tasks from offline data by learning a structured representation of the environment's temporal structure and using directional movements for policy "prompting".
https://openreview.net/forum?id=LhAuVPWq6q
Compressor summary: The paper proposes a resolution-invariant super-resolution method based on a hierarchical neural operator with self-attention and sinc filters, achieving better results than existing methods.
https://openreview.net/forum?id=Lgq1E92h1U
Compressor summary: Stacked Deep Sets and Quantile Pooling are novel methods for learning from set data that combine the strengths of max and average pooling and improve deep set networks.
https://openreview.net/forum?id=Lgh8bhWpVC
Compressor summary: The method generates 3D scenes by disentangling them into separate objects using a pretrained text-to-image model and multiple NeRFs, while ensuring the composited scenes resemble the original image.
https://openreview.net/forum?id=Lg8nw3ltvX
Compressor summary: The paper proposes a method to improve online continual learning by balancing new and old data in the optimization geometry, reducing instabilities and improving accuracy.
https://openreview.net/forum?id=Lfp5Dk1xb6
Compressor summary: The paper introduces a new token-based world model agent called REM that uses a Parallel Observation Prediction mechanism to speed up imagination and achieve superhuman performance on Atari 100K games.
https://openreview.net/forum?id=LfJgeBNCFI
Compressor summary: The authors propose DS-Agent, a framework that combines large language models and case-based reasoning to automate data science tasks, achieving high performance and cost efficiency.
https://openreview.net/forum?id=Ld255Mbx9F
Compressor summary: The text discusses how increasing the width of neural networks can decrease catastrophic forgetting, but shows that this relationship has diminishing returns and explores new widths to verify this empirically.
https://openreview.net/forum?id=Lc1HlMo77m
Compressor summary: The paper explores how to improve open-world generalization of vision-language models by leveraging weaker models and introduces three customized ensemble strategies for different scenarios.
https://openreview.net/forum?id=LbcNAIgNnB
Compressor summary: The paper proposes a policy gradient method for state entropy maximization in reinforcement learning with partial observations and belief states, addressing practical challenges in real-world applications.
https://openreview.net/forum?id=LbEB39lZqp
Compressor summary: The paper proposes USTAD, a unified Transformer model for information retrieval and ranking that uses novel distillation methods and asymmetric architectures to achieve high performance with fewer parameters.
https://openreview.net/forum?id=Lb8G2dZjcB
Compressor summary: This paper explains why inner product-based decoders struggle with graph data and proposes simple changes to improve them.
https://openreview.net/forum?id=LabSWooau0
Compressor summary: The paper proposes a new MAML-based optimizer that adapts PID control gains at each layer to improve learning efficiency and generalization across different tasks and domains.
https://openreview.net/forum?id=LZkhKZvhHs
Compressor summary: The paper proposes a new image quality assessment method that removes irrelevant and noisy features from upstream networks using an adversarial approach and data distillation.
https://openreview.net/forum?id=LZeixIvQcB
Compressor summary: TabLog is a novel method that adapts predictive models to a target domain using unlabeled data, by discretizing numerical features, modeling feature dependencies, and introducing a contrastive loss for distribution shift.
https://openreview.net/forum?id=LYpGLrC4oq
Compressor summary: The paper proposes a predictive dynamic fusion framework for multimodal learning that reduces generalization error and calibrates potential uncertainty in open environments.
https://openreview.net/forum?id=LWRI4uPG2X
Compressor summary: ECInstruct is a new instruction dataset that improves the performance of large language models (LLMs) in e-commerce by adapting them to specific tasks and products, resulting in better generalization across domains.
https://openreview.net/forum?id=LWD7upg1ob
Compressor summary: The Aug-PE algorithm generates differentially private synthetic text using only API access to a large language model without any finetuning.
https://openreview.net/forum?id=LVgT0ShxN5
Compressor summary: This paper proposes a framework called tnGPS that uses large language models to automatically discover new tensor network structure search algorithms, improving performance in high-dimensional representation tasks.
https://openreview.net/forum?id=LVF4P1NNwO
Compressor summary: The paper proposes a method to make In-Context Learning (ICL) in large language models explicit and permanent using bias terms, improving interpretability and efficiency over existing methods.
https://openreview.net/forum?id=LTifAl5bKb
Compressor summary: The paper proposes a new neural compression method based on Riemannian geometry that improves inference accuracy without needing extra data or fine-tuning.
https://openreview.net/forum?id=LRnXPxDksA
Compressor summary: The paper proposes Bayesian level-perfect MMR, an improved objective for unsupervised environment design in reinforcement learning, and introduces the ReMiDi algorithm that achieves this objective.
https://openreview.net/forum?id=LRkJwPIDuE
Compressor summary: VideoPoet is a transformer-based language model that can create realistic videos from various inputs and can be adapted for different video generation tasks.
https://openreview.net/forum?id=LO4xhXmFal
Compressor summary: DE-COP is a method to detect if copyrighted content was used in training language models by asking multiple-choice questions to the model, using BookTection as a benchmark.
https://openreview.net/forum?id=LM7j0zrUZB
Compressor summary: The paper studies how to learn an optimal policy for unknown tasks efficiently using meta reinforcement learning, and shows that strong identifiability assumptions enable faster regret minimization.
https://openreview.net/forum?id=LLdeUPOUXk
Compressor summary: The paper presents an efficient decentralized optimization method that uses different mini-batch sizes per node, has a sharper global condition number dependency, and requires fewer oracle calls and comparable communication cost than existing methods.
https://openreview.net/forum?id=LJcIIhqGDN
Compressor summary: SF-Gen is a novel reinforcement learning approach that leverages successor features for efficient and high-quality controllable text generation with multiple target subjects.
https://openreview.net/forum?id=LJ34pX1U5g
Compressor summary: The paper proposes a new method for causal inference with heterogeneous data that uses weighted propensity score models and federated learning to improve accuracy and privacy.
https://openreview.net/forum?id=LIQYhV45D4
Compressor summary: FLUTE is a novel federated representation learning algorithm with provable performance guarantees, data-independent initialization, and a designed objective function for under-parameterized linear models and beyond.
https://openreview.net/forum?id=LIPGadocTe
Compressor summary: FedLCB-Q is a Q-learning variant for federated offline RL that achieves linear speedup and communication efficiency by collaboratively leveraging offline datasets at multiple agents.
https://openreview.net/forum?id=LHGMXcr6zx
Compressor summary: EfficientZero V2 is a generalized framework for sample-efficient reinforcement learning that performs better than DreamerV3 in various tasks and domains.
https://openreview.net/forum?id=LH6R06NxdB
Compressor summary: The paper introduces Stepwise ORMs, which use synthetic data to improve reasoning refinement in language models, and shows that combining global and local refinements achieves better results than existing methods on GSM8K.
https://openreview.net/forum?id=LGz7GaUSEB
Compressor summary: The paper proposes a novel hierarchical reinforcement learning framework for optimizing multiplier design in integrated circuits, achieving better Pareto-optimal solutions with high sample efficiency and generalization.
https://openreview.net/forum?id=LGhtl9ktop
Compressor summary: The authors propose a dynamic neural generation network that skips or reduces computation for some data instances to accelerate inference without sacrificing accuracy in various NLP tasks.
https://openreview.net/forum?id=LGDYsBslWi
Compressor summary: This paper studies theoretical aspects of kernel-based statistical learning on distributional inputs, proving oracle inequalities and a generalization result for different embedding methods.
https://openreview.net/forum?id=LDq1JPdc55
Compressor summary: The authors propose using copyright traps with fictitious entries to detect the use of copyrighted content in large language models, especially those that do not naturally memorize.
https://openreview.net/forum?id=LCTmppB165
Compressor summary: CaM is a technique that adaptively merges caches to preserve critical token information and improve the performance of memory-efficient Large Language Models.
https://openreview.net/forum?id=LALSZ88Xpx
Compressor summary: The authors propose using non-local scoring rules like Brier and Spherical score for language generation, which can improve the quality of generated text across different models.
https://openreview.net/forum?id=L8nSGvoyvb
Compressor summary: Relaxed Quantile Regression improves prediction intervals by removing arbitrary constraints and allowing better coverage of skewed distributions.
https://openreview.net/forum?id=L6SRXG92s6
Compressor summary: The paper proposes a novel graph clustering method that uses structural information and does not need predefined cluster numbers, outperforming existing methods.
https://openreview.net/forum?id=L4ERlHrJRT
Compressor summary: Key points: - The challenge is to ensure DNNs use correct input features and avoid spurious correlations - Existing methods remove both correct and spurious features, leading to wrong interpretations and poor performance - A new iterative algorithm separates spurious from main-task concepts by estimating two orthogonal subspaces of the neural network representation - The algorithm outperforms existing methods on benchmark datasets from computer vision and natural language processing Summary: The paper proposes a novel algorithm that accurately identifies and removes spurious features from DNNs, improving interpretability and performance.
https://openreview.net/forum?id=L1eJ3NKPCd
Compressor summary: The paper investigates how well Transformers can learn and combine various language capabilities using autoregressive training and experiments on different composition methods.
https://openreview.net/forum?id=L1W9ZWPq9E
Compressor summary: The text introduces new compression methods for summarizing target distributions using biased input sequences and shows their effectiveness in various applications.
https://openreview.net/forum?id=L0VoOdjCUb
Compressor summary: This paper surveys rotation representations for machine learning models and provides guidance on choosing suitable ones based on their properties, input and output locations, and angle size.
https://openreview.net/forum?id=L057s2Rq8O
Compressor summary: The paper explores KV cache quantization for large language models and proposes a 2-bit algorithm called KIVI that significantly reduces memory usage and improves batch size, speed, and quality.
https://openreview.net/forum?id=KzACYw0MTV
Compressor summary: Quest is a query-aware algorithm that speeds up self-attention in long-context LLMs by selecting the most critical KV cache pages for attention, achieving significant speedup and accuracy.
https://openreview.net/forum?id=KycvgOCBBR
Compressor summary: The paper proposes a new method called GIC that accurately infers group labels, improving the worst-group performance of ERM models by exploiting properties of spurious correlations and semantic consistency.
https://openreview.net/forum?id=KwgAThfxEd
Compressor summary: The paper studies a modified gradient descent algorithm that uses local curvature information and achieves optimal computational complexity for singular statistical models.
https://openreview.net/forum?id=KviM5k8pcP
Compressor summary: The paper proposes and tests AI control techniques to prevent powerful but untrusted LLMs from causing harmful outcomes even if they try to subvert safety measures.
https://openreview.net/forum?id=Kt4fwiuKqf
Compressor summary: The paper proposes PGPS-LwS, a novel particle-based Bayesian inference method that uses a Neural Network-learned vector field to guide particles from an initial distribution to the target distribution along a Log-weighted Shrinkage density path, which improves accuracy and calibration in synthetic and real-world tasks.
https://openreview.net/forum?id=KsUddQl39v
Compressor summary: The paper proposes PART, a method that adjusts the perturbation budget for each pixel based on its importance, improving accuracy and robustness in adversarial training.
https://openreview.net/forum?id=Kqa5JakTjB
Compressor summary: Powder is a new FCL algorithm that uses prompts to facilitate positive knowledge transfer across tasks and clients while reducing communication costs and privacy concerns.
https://openreview.net/forum?id=KpeGdDzucX
Compressor summary: PSB is a new, efficient, and stable slot learning architecture for sequential inputs that improves object-centric scene decomposition and understanding.
https://openreview.net/forum?id=KpUdNe9lsr
Compressor summary: Key points: - Graph representation helps in MARL by capturing correlations among entities - HGAP network uses graph attention to enforce PI and PE properties - HGAP is effective, efficient, adaptable, and transferable across MARL benchmarks Summary: HGAP is a novel graph-based policy network that enables agents in multi-agent reinforcement learning to learn from correlated entities with permutation equivariant properties.
https://openreview.net/forum?id=KmCoS6WkgG
Compressor summary: The paper presents an efficient autoregression-based vision model that achieves proficiency in various visual tasks with a reduced parameter footprint and training data requirements.
https://openreview.net/forum?id=Kjww7ZN47M
Compressor summary: MathScale is a method that uses large language models to create high-quality math data, improving their mathematical reasoning abilities and achieving state-of-the-art performance on MWPBench.
https://openreview.net/forum?id=KjazcKPMME
Compressor summary: The paper explores how iterative prompting can improve the accuracy and truthfulness of large language models, introducing new variants that address previous challenges.
https://openreview.net/forum?id=KgfGxXbjjE
Compressor summary: This paper introduces CKGConv, a novel and general graph convolution framework that uses continuous functions of pseudo-coordinates derived via graph positional encoding, which is flexible, expressive, and performs well on various graph datasets.
https://openreview.net/forum?id=KfXXPCcobh
Compressor summary: SSD is a pre-training method that adaptively switches between sparse and dense training, improving efficiency and inference speed for Transformers.
https://openreview.net/forum?id=KfN76nAcOO
Compressor summary: HND uses a structured hierarchy of known and novel classes to detect fine-grained novelty, separating it from known classes with a unique loss function and evidence margin.
https://openreview.net/forum?id=Kf9CqdI8Rb
Compressor summary: The paper proposes a new neural decoder for binary linear block codes, which improves the efficiency of data transfer over noisy channels and outperforms existing methods.
https://openreview.net/forum?id=Kbd9A4lVoX
Compressor summary: FedPIN is a personalized federated learning method that uses causal models to distinguish between personalized and spurious features, improving out-of-distribution generalization.
https://openreview.net/forum?id=KaAQu5rNU1
Compressor summary: MolCRAFT is a new SBDD model that generates stable and high-affinity molecules by operating in the continuous parameter space and using a noise reduced sampling strategy.
https://openreview.net/forum?id=KYrAZSbEv6
Compressor summary: The text proposes a generative network model to understand the statistical foundation of using IPF for inferring dynamic networks from time-aggregated adjacency matrices and time-varying marginals, and introduces an algorithm to improve convergence on sparse data.
https://openreview.net/forum?id=KXsUCgn9Ks
Compressor summary: The paper proposes BEB, a theoretical approach to investigate alignment limitations in large language models, showing that any partial alignment is vulnerable to adversarial prompting attacks, which have been demonstrated experimentally with chatGPT jailbreaks.
https://openreview.net/forum?id=KVvku47shW
Compressor summary: The paper investigates how neural scaling laws change with synthetic data in training and explores various decay phenomena that may lead to model collapse.
https://openreview.net/forum?id=KVa4i4RR1O
Compressor summary: The authors introduce *convSeq*, an unsupervised method that uses backpropagation to optimize spatiotemporal filters for efficiently detecting repetitive patterns in large neural recordings, potentially improving our understanding of neural circuit function.
https://openreview.net/forum?id=KU9mn6deDR
Compressor summary: UPAM is a novel framework that investigates the robustness of T2I models from the attack perspective, deceiving both textual and visual defenses using gradient-based optimization and ensuring attack stealthiness.
https://openreview.net/forum?id=KSNl7VgeVr
Compressor summary: Premier-TACO is a multitask feature representation learning method that improves efficiency in sequential decision-making tasks by using minimal expert demonstrations and incorporating a novel negative example sampling strategy for temporal action contrastive learning.
https://openreview.net/forum?id=KOW9ncAiRo
Compressor summary: This paper studies kernel quantile regression with random features (KQR-RF), improves its error decomposition, connects it to kernel ridge regression with random features (KRR-RF), and provides optimal learning rates under various conditions.
https://openreview.net/forum?id=KOTutrSR2y
Compressor summary: MM-Vet is a benchmark for evaluating large multimodal models based on their integration of core vision-language capabilities, with a unified scoring metric that works across question types.
https://openreview.net/forum?id=KNedb3bQ4h
Compressor summary: The paper proposes a single-layer online learning algorithm with wavelet detection and adaptive restart for minimizing dynamic regret in non-stationary settings, outperforming two-layer ensembles.
https://openreview.net/forum?id=KLmWRMg6nL
Compressor summary: FairGrad is a novel optimization objective for multi-task learning that maximizes utility and ensures fair resource allocation among tasks, improving performance in supervised and reinforcement learning.
https://openreview.net/forum?id=KJhLpzqNri
Compressor summary: EP-GFlowNet is a method to sample from large product distributions efficiently using local GFlowNets and a global model learned with a novel aggregating balance condition, enabling parallel and federated Bayes.
https://openreview.net/forum?id=KJL2b6BthC
Compressor summary: The Algorithm of Thoughts is a novel strategy that boosts LLMs' reasoning capacities by guiding them through algorithmic reasoning pathways using few queries, outperforming other methods and suggesting LLMs can weave their intuition into optimized searches.
https://openreview.net/forum?id=KI3JKFKciG
Compressor summary: DFD is a model compression technique that adapts distillation constraints based on the inconsistency in disparity between teacher and student feature maps, improving object detection performance.
https://openreview.net/forum?id=KHymcy2xxF
Compressor summary: Our method generates synthetic training data for image classification based on each class's actual data needs, improving performance on imbalanced and balanced datasets.
https://openreview.net/forum?id=KCVCFsPkrm
Compressor summary: The paper presents a tighter privacy analysis for noisy gradient descent and its variants using the framework of f-differential privacy and a new construction called shifted interpolated processes, which works for various optimization and batch settings.
https://openreview.net/forum?id=KB6slOUQP9
Compressor summary: This paper introduces training-free algorithms to accelerate diffusion models' sampling, achieving faster convergence rates than existing methods.
https://openreview.net/forum?id=K9NTPRvVRI
Compressor summary: The paper investigates how adding new knowledge to large language models affects their existing knowledge and introduces a metric and a framework to mitigate this perturbation.
https://openreview.net/forum?id=K6xxnKN2gm
Compressor summary: The study examines how large language models are brittle and unsafe, especially when pruned or modified, and suggests the need for better safety strategies.
https://openreview.net/forum?id=K6HpbvkrwO
Compressor summary: The study proposes an adaptive experiment that optimizes both covariate density and propensity score to efficiently estimate average treatment effects with lower asymptotic variance.
https://openreview.net/forum?id=K5h6VAsJaV
Compressor summary: The GGNS algorithm is a powerful tool for scientific computing that uses advanced techniques to efficiently sample complex probability distributions and estimate their properties.
https://openreview.net/forum?id=K3fEkECWgu
Compressor summary: VoxBind is a new model that generates 3D molecules based on protein structures using atomic density grids and voxel-denoising networks, resulting in more diverse, less clashing, and higher-affinity molecules than existing methods.
https://openreview.net/forum?id=JzWFmMySpn
Compressor summary: ExcessMTL is a multi-task learning method that balances tasks by their distances to convergence, improving performance under label noise.
https://openreview.net/forum?id=JymXv7mkrQ
Compressor summary: The paper proposes a method called WCA that uses localized visual prompting and cross-alignment of visual and textual descriptions to improve zero-shot image classification performance.
https://openreview.net/forum?id=Jvh8HM9YEJ
Compressor summary: MH-pFLID is a new federated learning method for medical applications that uses a lightweight messenger model, reduces biased data collection, and does not require public datasets or extra resources.
https://openreview.net/forum?id=JvMLkGF2Ms
Compressor summary: The paper explores guardrails as a key technology for mitigating risks associated with large language models and suggests a systematic approach based on socio-technical methods and multi-disciplinary collaboration.
https://openreview.net/forum?id=JtkruFHcRK
Compressor summary: DAEDL is a novel method for improving uncertainty estimation and classification in deep learning by integrating feature space density with output from EDL and using a new parameterization.
https://openreview.net/forum?id=Jtjurj7oIJ
Compressor summary: The paper critiques the use of robotic simulations for real-world manipulation tasks, arguing that scaling simulators is not enough to achieve human-compatible general-purpose systems.
https://openreview.net/forum?id=JsPvL6ExK8
Compressor summary: The paper introduces Prometheus, a new dataset for fluid dynamics modeling, and proposes DGODE, a method that learns disentangled representations to improve out-of-distribution generalization.
https://openreview.net/forum?id=JpzIGzru5F
Compressor summary: The authors propose a novel way to represent causal models without DAGs, design a two-stage generative model that infers causal order from data, and use a new attention mechanism to capture causality in transformer-based architecture.
https://openreview.net/forum?id=JndWnomyIc
Compressor summary: The text proposes a framework to transform in-processing techniques for group fairness into post-processing methods, which can be applied to more problem settings and preserve or improve fairness-error trade-offs.
https://openreview.net/forum?id=JnA9IveEwg
Compressor summary: MGSE is a framework that improves graph self-supervised learning by using multiple student models to learn from a single teacher model, capturing multi-granular knowledge and achieving better generalization abilities.
https://openreview.net/forum?id=JcxlFe2fGC
Compressor summary: The paper introduces TINT, an efficient construction that allows transformers to simulate and fine-tune more complex models during inference, improving performance on language modeling and downstream tasks.
https://openreview.net/forum?id=JYcbgiSh0L
Compressor summary: The paper proposes a stochastic optimization algorithm that achieves optimal convergence rates under general recurrent data sampling schemes, even for non-convex and non-smooth problems with constraints.
https://openreview.net/forum?id=JVhUR8q27o
Compressor summary: A-BAD-BO is a novel algorithm that uses Bayesian optimization to jointly train ML components in complex systems, improving system performance and efficiency.
https://openreview.net/forum?id=JVORowD4MD
Compressor summary: The paper proposes a variant of sequential importance sampling that combines its efficiency with accuracy guarantees from rejection sampling, resulting in faster estimations of high-dimensional integrals such as the permanent of nonnegative matrices.
https://openreview.net/forum?id=JV84NVo1em
Compressor summary: The paper introduces Adaptation Safety, a novel concept for opponent exploitation in games, and proposes the Opponent eXploitation Search (OX-Search) framework that improves safety and robustness in online poker games.
https://openreview.net/forum?id=JUa5XNXuoT
Compressor summary: The paper proposes a transformer model with discrete bottlenecks that can learn compressed representations of observations and actions, enabling it to extract interpretable cognitive maps for path planning in partially observed environments.
https://openreview.net/forum?id=JU3xHh1vWw
Compressor summary: This paper studies how to recover true relevance for ranking models from biased click logs by analyzing graph connectivity and proposing node intervention and node merging methods.
https://openreview.net/forum?id=JSYN891WnB
Compressor summary: The authors propose Text-Aided Clustering (TAC), a method that uses WordNet nouns to enhance image features and cross-modal neighborhood information distillation for image clustering, achieving state-of-the-art results on various benchmarks.
https://openreview.net/forum?id=JQlEUfzhuA
Compressor summary: The paper compares two paradigms for learning from human preferences, RLHF and DPO, in different settings and provides minimax bounds and convergence rates.
https://openreview.net/forum?id=JPNBFWQ9H2
Compressor summary: Bifurcated attention is a method that reduces memory IO costs in language model inference by dividing the attention mechanism into two GEMM operations, improving efficiency and latency for high batch sizes and long context lengths.
https://openreview.net/forum?id=JOrLz5d7OW
Compressor summary: The ProtoFormer framework uses prototype learning to understand and represent motion patterns in videos for tasks like optical flow, scene depth, object tracking, and video stabilization.
https://openreview.net/forum?id=JObct1zyTb
Compressor summary: The text introduces FRGR, a framework that helps improve neural logic machines' reasoning and decision-making abilities by identifying and penalizing repeated mistakes during training.
https://openreview.net/forum?id=JOKOsJHSao
Compressor summary: The text introduces information-directed pessimism, a novel method for offline reinforcement learning that uses Stein discrepancy to estimate and account for distribution mismatch between batch data and current policy.
https://openreview.net/forum?id=JNeeRjKbuH
Compressor summary: The paper proposes a private algorithm to improve the fairness of regressors by remapping their outputs based on output distributions' estimated barycenter.
https://openreview.net/forum?id=JNN6QHhLHB
Compressor summary: IF-COMP is a new method for measuring uncertainty in models that considers different labels and works well on various tasks related to reliability and calibration.
https://openreview.net/forum?id=JNHK11bAGl
Compressor summary: FCSRL is a novel framework for safe reinforcement learning that combines representation learning with feasibility-oriented objectives to improve policy learning and constraint estimation.
https://openreview.net/forum?id=JKPhWzp7Oi
Compressor summary: The paper analyzes D-SGD's generalization error using algorithmic stability and shows that the choice of communication graph can affect performance, sometimes improving it.
https://openreview.net/forum?id=JJpOssn0uP
Compressor summary: Prodigy is a new algorithm that estimates the distance to the solution in adaptive learning methods, improving upon existing methods by a factor of $\mathcal{O}(\sqrt{\log(D/d_0)})$ and achieving test accuracy close to hand-tuned Adam.
https://openreview.net/forum?id=JJZBZW28Gn
Compressor summary: SDCD is a new method for inferring causal relationships from data that improves stability, speed, and scalability over existing DCD methods.
https://openreview.net/forum?id=JJSj8UXqd4
Compressor summary: The article addresses two open questions about lifted multicut polytopes and answers one of them while showing the difficulty of answering the other.
https://openreview.net/forum?id=JIWtKcR78C
Compressor summary: The paper studies how humans make generalizations about large language models' performance across different tasks and shows that more capable models may underperform in high-stakes situations due to misalignment with human expectations.
https://openreview.net/forum?id=JHRvP84SQ5
Compressor summary: The paper develops a general framework for unbounded concentration inequalities for heavy-tailed distributions and applies it to statistical learning theory problems.
https://openreview.net/forum?id=JGL39NaARS
Compressor summary: The MFTN method improves HISR by extracting multi-scale features from LR-HSI and HR-MSI using Transformers and aggregating them to create a high-quality HR-HSI.
https://openreview.net/forum?id=JD03zxWZzs
Compressor summary: FedRed is a new federated learning method that reduces communication costs while maintaining performance by using doubly regularized drift correction.
https://openreview.net/forum?id=JCG0KTPVYy
Compressor summary: COFT is a method that reduces hallucination by highlighting key texts of different granularity levels using recaller, scorer, and selector components.
https://openreview.net/forum?id=JBaPBPrn93
Compressor summary: Attention layers are more effective than random features for NLP tasks because they have high word sensitivity, allowing them to capture contextual meaning better.
https://openreview.net/forum?id=JApt4Ty89Y
Compressor summary: The paper investigates quantum advantages for online optimization problems and proposes quantum online quasi-Newton methods that achieve better regret than classical algorithms.
https://openreview.net/forum?id=JAfIDm7NED
Compressor summary: MAAD is a sample-efficient imitation learning algorithm that uses a surrogate reward signal and an inverse dynamics model to learn from expert observations in various environments.
https://openreview.net/forum?id=JA6ThxAmth
Compressor summary: The text discusses the limitations of concept-based explainability methods for deep learning systems and proposes a new algorithm that incorporates inter-concept relationships to enhance explanations.
https://openreview.net/forum?id=J9YKDvqr65
Compressor summary: The paper proposes a lookahead mechanism for SAM to improve convergence stability and efficiency in finding flatter minima for adversarial training.
https://openreview.net/forum?id=J6prHJsIlf
Compressor summary: FANS is a new method for explaining machine learning models that uses perturbation tests and counterfactual reasoning to determine feature importance, which works better than existing methods on six benchmarks.
https://openreview.net/forum?id=J5Yg7HMy39
Compressor summary: The proposed method clusters functional data from a Gaussian process mixture using multiple one-dimensional projections and an ensemble of Gaussian mixture models, achieving faster computation and better performance than existing methods.
https://openreview.net/forum?id=J5VB1h3Aed
Compressor summary: The paper studies how to use generative vision-language models for image-text retrieval tasks and proposes a method to reduce linguistic bias in the model's output.
https://openreview.net/forum?id=J4LTDgwAZq
Compressor summary: The paper derives optimal robust Bellman operators for interconnected uncertainties in MDPs, leading to faster robust value iteration methods and revealing novel threshold behavior and policy resilience properties.
https://openreview.net/forum?id=J4HJUF70qm
Compressor summary: The paper proposes a new Clustered Federated Learning algorithm that groups clients based on their model updates, leading to better clustering and learning performance.
https://openreview.net/forum?id=J3xYTh6xtL
Compressor summary: Label shift correction (LSC) estimates and adjusts the test label distribution to reduce generalization error in real-world applications with non-uniform label distributions.
https://openreview.net/forum?id=J1NIXxiDbu
Compressor summary: The text introduces PANDA, a new message passing method for GNNs that expands the width of highly central nodes to prevent over-squashing and improve long-range information propagation without distorting graph topology.
https://openreview.net/forum?id=J16WEPdqhJ
Compressor summary: The paper proposes a faster and more practical policy gradient method for robust Markov decision processes with environmental perturbations and large state spaces.
https://openreview.net/forum?id=J0ty1o7nCj
Compressor summary: The paper proposes using deep learning to learn efficient preconditioners for a linear solver, allowing for faster and more accurate solutions to partial differential equations across different resolutions.
https://openreview.net/forum?id=Izv7gBnap3
Compressor summary: Federated robust averaging (FedRo) can converge well for smooth non-convex loss, but its improvement rate depends on client subsampling and local steps.
https://openreview.net/forum?id=IzqpUC34Jg
Compressor summary: The paper proposes a framework to assess the privacy cost of data-dependent pre-processing in differentially private machine learning pipelines and provides explicit privacy guarantees for several algorithms.
https://openreview.net/forum?id=IyeXM58vIC
Compressor summary: The paper introduces Total Variation Floodgate, a method to measure variable importance in classification problems without model assumptions, and provides algorithms to infer it using regression functions and confidence bounds.
https://openreview.net/forum?id=IxZ4xaHSYG
Compressor summary: This paper studies how different smoothing distributions affect Randomized Smoothing's robustness against adversarial examples and shows that Exponential General Gaussian distribution improves the defense mechanism.
https://openreview.net/forum?id=IwqE4QqBew
Compressor summary: The paper introduces an r times r preconditioner for Low-Rank Adaptation (LoRA), which improves convergence, reliability, and robustness of fine-tuning with minimal overhead.
https://openreview.net/forum?id=IuvpVcGUOB
Compressor summary: The paper proposes PCLP, a novel SSMLL method that infers label correlations using SCM to enhance pseudo-labeling and improve performance.
https://openreview.net/forum?id=Irkcamqg4d
Compressor summary: The paper presents BDMatch, a new method for open-set semi-supervised learning that uses binary decomposition to address class-imbalance and representation-compromise issues.
https://openreview.net/forum?id=IpSKpOY2EH
Compressor summary: The paper proposes methods to reduce memory overhead in fine-tuning large models using Approximate Backpropagation theory and novel activation functions.
https://openreview.net/forum?id=IpPnmhjw30
Compressor summary: The meta-continual learning framework combines neural networks' representational power and statistical models' robustness to forgetting by using sequential Bayesian update rules and meta-learning.
https://openreview.net/forum?id=IoUOhnCmlX
Compressor summary: The paper studies likelihood-based methods for recovering complex signals from multiple measurements with noise, provides a new MSE upper bound, and introduces bagged Deep Image Priors combined with projected gradient descent and Newton-Schulz algorithm to achieve state-of-the-art performance.
https://openreview.net/forum?id=InUUQkExsw
Compressor summary: We develop two new risk-sensitive offline reinforcement learning algorithms for linear MDPs that use entropic risk and improve upon existing methods.
https://openreview.net/forum?id=IgwtflILyj
Compressor summary: Bloop is a method for improving deep learning estimation pipelines by blending auxiliary loss gradients with training loss gradients using orthogonal projection and moving average, leading to better performance on NLP and vision tasks.
https://openreview.net/forum?id=If6Q9OYfoJ
Compressor summary: LiRE is a novel approach for offline PbRL that uses second-order preference information from human feedback to learn reward models more effectively than existing methods, as shown by experiments on a new dataset.
https://openreview.net/forum?id=If4xW9vF7U
Compressor summary: DCA is a training-free method that improves LLMs' ability to handle long context sequences by using chunk-based attention modules.
https://openreview.net/forum?id=IejxxE9DO2
Compressor summary: DySymNet is a novel neural-guided dynamic symbolic network that uses reinforcement learning to optimize the structure of symbolic networks for discovering mathematical expressions from data.
https://openreview.net/forum?id=IckJCzsGVS
Compressor summary: Proteus is a new deep diffusion network that can create novel proteins with efficient and effective designability, without depending on pre-trained structure prediction networks.
https://openreview.net/forum?id=IaV6AgrTUp
Compressor summary: Implicit representations enable deep learning models to enforce geometric constraints in image segmentation by mapping coordinates to pixel properties.
https://openreview.net/forum?id=IYI61L7SPk
Compressor summary: The paper proposes algorithms for sequential decision-making with rare policy switch and batch learning constraints, achieving sub-linear regret under various function classes.
https://openreview.net/forum?id=IWi6iLZeRG
Compressor summary: The paper compares deeper and wider neural networks in terms of generalization error, influenced by factors such as sample points, parameters, and loss function regularity, and applies the theory to partial differential equations using deep Ritz and PINN methods.
https://openreview.net/forum?id=IW45Dr1Kxi
Compressor summary: The paper proposes new positional encodings for graph neural networks using quantum computers, which can improve the performance of state-of-the-art models on standard benchmarks and large-scale datasets.
https://openreview.net/forum?id=IUijgjJgWO
Compressor summary: The paper analyzes the robustness of large language and vision-language models by showing their vulnerability to adversarial permutation in multiple-choice question answering.
https://openreview.net/forum?id=IUBhvyJ9Sr
Compressor summary: HIEROS is a hierarchical policy that improves sample efficiency in DRL by learning time abstracted world representations and imagining trajectories at multiple time scales, achieving better performance and exploration than existing methods.
https://openreview.net/forum?id=ISG3l8nXrI
Compressor summary: QDAC is an actor-critic deep reinforcement learning algorithm that learns high-performing and diverse behaviors, enabling better adaptation to new situations.
https://openreview.net/forum?id=INb8xV1xmf
Compressor summary: The paper proposes a novel framework called SP-GS that uses 3D Gaussians and superpoints to render dynamic scenes with high quality and fast speed.
https://openreview.net/forum?id=IGdpKP0N6w
Compressor summary: The paper investigates how neural networks learn low-order moments of data distributions and shows that they can perform well on maximum-entropy distributions with similar statistics, but lose this ability later in training. The authors also extend the distributional simplicity bias to discrete domains and use optimal transport methods to edit sample statistics.
https://openreview.net/forum?id=ICvWruTEDH
Compressor summary: The paper presents a new graph signal denoising method called Graph Adversarial Diffusion Convolution (GADC), which improves robustness and performance on heterophilic graphs by using a min-max optimization formulation.
https://openreview.net/forum?id=IC9UZ8lm25
Compressor summary: The text introduces MultiMax, a new function that improves the interpretability of machine learning algorithms by adaptively suppressing irrelevant entries and preserving multi-modality.
https://openreview.net/forum?id=IArWwIim8M
Compressor summary: The paper proposes a new method to optimize ReLU networks' inputs by considering activation patterns changes, using differentiable representations and regularization terms for better local descent properties and improved performance in various tasks.
https://openreview.net/forum?id=I4HTPws9P6
Compressor summary: The paper analyzes the training dynamics and generalization capabilities of Transformers for in-context learning, studying how different factors and components affect performance and exploring the effects of model pruning.
https://openreview.net/forum?id=I44Em5D5xy
Compressor summary: The paper proposes Molecular Conformer Fields (MCF), a diffusion generative model that predicts molecular conformers by learning a distribution over functions mapping elements from a molecular graph to their 3D positions, achieving state of the art results with simplicity and scalability.
https://openreview.net/forum?id=Hzpt1Gws9g
Compressor summary: NEM-U is a framework that uses a masking network to provide fast and accurate explanations for learned vector embeddings in unsupervised representation learning.
https://openreview.net/forum?id=Hy88Jp0kQT
Compressor summary: This paper studies how the distribution of preference datasets affects the learning dynamics and behavior of large language models aligned to human intentions, providing theoretical insights and empirical validation.
https://openreview.net/forum?id=HwVZbPbMjw
Compressor summary: The paper introduces passive RL, a method to convert passive observations into actionable insights, and proposes MSCP, a novel algorithm that leverages two planners at distinct scales to improve online RL with passive data.
https://openreview.net/forum?id=HvwOtYzHBX
Compressor summary: LLark is an instruction-tuned multimodal model for understanding music that uses a generative music model and a language model, trained on open-source data, and performs well on various tasks.
https://openreview.net/forum?id=HusShERjlc
Compressor summary: Key points: - Model ensemble is a trending method for improving Out-of-Distribution (OOD) detection performance by expanding feature representation field - Previous ensemble methods have limited variability in weights and diversity - Multi-Comprehension Ensemble uses various supervision tasks to form different comprehensions of data and labels, resulting in more diverse feature representation - MC Ensemble outperforms naive Deep Ensemble and standalone model in OOD detection task Summary: The authors propose a novel ensemble method called Multi-Comprehension Ensemble that uses various supervision tasks to create diverse feature representations for better Out-of-Distribution detection than previous methods.
https://openreview.net/forum?id=Htw0bSgjXE
Compressor summary: CurBench is a new benchmark for curriculum learning that evaluates machine learning models across 3 research domains, 3 settings, and 2 dimensions of performance and complexity.
https://openreview.net/forum?id=Ht20wtgaty
Compressor summary: The paper introduces CTMA, a meta-aggregator for Byzantine-robust training in distributed ML systems, and proposes a gradient estimation technique based on double-momentum strategy with theoretical and practical advantages.
https://openreview.net/forum?id=HsseRq2FAx
Compressor summary: The paper proposes Dr. Strategy, an MBRL agent with a novel dreaming strategy based on spatial divide-and-conquer, which improves performance in complex navigation tasks.
https://openreview.net/forum?id=HssOwuZiaB
Compressor summary: The authors study abrupt improvements in transformer training loss when facing multi-step tasks, called Eureka-moments, and propose ways to improve training by addressing issues in the Softmax function in self-attention blocks.
https://openreview.net/forum?id=HsliOqZkc0
Compressor summary: The study shows that different diffusion models often produce similar outputs due to learning distinct distributions influenced by training data size, and this property holds for various model variants.
https://openreview.net/forum?id=Hs9GcILuZN
Compressor summary: FLIPAD is a new method for identifying whether a sample was generated by a specific generative model or not, based on final-layer inversion and anomaly detection, which works in the open-world setting and is efficient and flexible.
https://openreview.net/forum?id=HrzQZXzrN2
Compressor summary: The paper proposes a method to compare predictive models with existing decision policies by identifying and ignoring regions of uncertainty in the data, and applies it to evaluate a healthcare policy modification.
https://openreview.net/forum?id=HpT19AKddu
Compressor summary: The paper proposes CCPG, a coarser representation of hidden causal graphs that can be learned with fewer CI tests than existing algorithms, especially when the causal graph is fully identifiable.
https://openreview.net/forum?id=HmKMpJXH67
Compressor summary: SLWI is a novel online test-time adaptation framework that uses Bayesian filtering to continually update model weights and adapt to target domain changes, reducing errors and improving performance in various distribution shift scenarios.
https://openreview.net/forum?id=HkWxjpUV0S
Compressor summary: The paper introduces Object-Oriented CDM (OOCDM), a model for learning causal dependencies among objects in large-scale environments, which improves on existing CDMs in various aspects.
https://openreview.net/forum?id=HkCRgoGtt6
Compressor summary: The paper introduces a new benchmark and model for long-context retrieval in machine learning systems, addressing challenges like evaluation, pretraining, and finetuning with limited GPU memory.
https://openreview.net/forum?id=Hjwx3H6Vci
Compressor summary: The paper proposes a distribution alignment optimization method (DisA) that improves deep neural networks' performance on long-tailed datasets by inducing the Neural Collapse phenomenon.
https://openreview.net/forum?id=Hh8pUBfxXh
Compressor summary: The paper proposes MMPareto algorithm to address gradient conflict in multimodal learning by using Pareto integration and achieving better generalization with unimodal assistance.
https://openreview.net/forum?id=Hg7C5YYifi
Compressor summary: This paper proposes a novel method for continual graph neural architecture search, which enables learning new tasks without forgetting past knowledge and addressing architecture conflicts.
https://openreview.net/forum?id=HfxFasUfbN
Compressor summary: The paper proposes a new low-bit integer training framework that allows adaptive mixed-precision allocation for weights, activations, and gradients, achieving significant BitOPs reduction and minimal performance loss in various tasks.
https://openreview.net/forum?id=HbdeEGVfEN
Compressor summary: The text discusses how representation topologies for classification and regression are different, and proposes a regularizer called PH-Reg that aligns the feature space with the target space in regression tasks based on the Information Bottleneck principle.
https://openreview.net/forum?id=HaBVzgSdM7
Compressor summary: The paper presents DAT, an end-to-end model that unifies scene text detection, layout analysis, and document page detection with across-granularity interactive attention and prompt-based segmentation modules, achieving state-of-the-art results on various text-related tasks.
https://openreview.net/forum?id=HZyOz9VEg4
Compressor summary: The paper proposes geometric pruning for dynamical systems reconstruction, which reduces parameter load while preserving performance by generating specific network topologies that work well with recurrent neural networks (RNNs).
https://openreview.net/forum?id=HZ6lrZzB02
Compressor summary: The paper proposes a bidding system based on game theory to estimate causal effects in RCTs with competing treatments, and shows that it has a pure Nash equilibrium that minimizes estimation error.
https://openreview.net/forum?id=HUJK9dFOW6
Compressor summary: The paper presents differentiable DRO layers for mixed-integer problems that enable decision-focused learning by embedding ambiguity sets as a layer, using dual-view methodology and importance sampling to estimate gradients.
https://openreview.net/forum?id=HTNgNt8CTJ
Compressor summary: The paper explores how adding subgraph information and using margin theory can improve the generalization performance of graph isomorphism algorithms, such as 1-WL, without necessarily increasing their expressivity.
https://openreview.net/forum?id=HTMFUKAm8B
Compressor summary: This paper compares algorithms for managing budget pacing and return-on-spend constraints in internet advertising, finding that a min-pacing algorithm performs well and is more coordinated than a sequential algorithm.
https://openreview.net/forum?id=HQtTg1try7
Compressor summary: Key points: - The paper revisits the problem of making image classifiers robust to small perturbations and develops scaling laws for adversarial training. - The paper finds inefficiencies in prior art and proposes a compute-efficient setup that surpasses previous methods with fewer FLOPs. - The paper also predicts a plateau in robustness at around 90% due to the generation of invalid images by adversaries, which is consistent with human performance. Summary: The paper analyzes and improves adversarial training for image classifiers, finding trade-offs between efficiency and robustness, and revealing the limits of perfect accuracy.
https://openreview.net/forum?id=HPXRzM9BYZ
Compressor summary: The paper introduces a framework to predict models' Out-of-Distribution performance using in-distribution measurements and shows that Lowest Common Ancestor distance can explain why Visual-Language Models generalize better than Vision Models.
https://openreview.net/forum?id=HPQaMmABgK
Compressor summary: Stochastic value-based RL methods consider a sublinear number of actions in each iteration, reducing computational burden and improving performance in complex environments.
https://openreview.net/forum?id=HPLzSCOecY
Compressor summary: The paper proposes LARA, a method for handling multiple time-constrained learning tasks with limited resources by predicting progress, allocating resources adaptively, and balancing errors.
https://openreview.net/forum?id=HOoVTsPPn7
Compressor summary: Orthogonal Bootstrap is a new method for reducing computational cost and improving accuracy in uncertainty simulations by decomposing the target into two parts.
https://openreview.net/forum?id=HOMXUneCTR
Compressor summary: The text discusses how transformers are biased towards symmetric functions and uses representation theory to analyze their performance on symmetric datasets like WikiText.
https://openreview.net/forum?id=HOG80Yk4Gw
Compressor summary: RACoon is a scalable relational verifier for deep neural networks that uses cross-execution dependencies to precisely verify properties like robustness and hamming distance.
https://openreview.net/forum?id=HO0g6cHVZx
Compressor summary: EiG-Search is a training-free approach to generate efficient and comprehensive subgraph-level explanations for Graph Neural Networks using an edge-induced search algorithm based on gradient importance.
https://openreview.net/forum?id=HLHQxMydFk
Compressor summary: The paper proposes a novel algorithm for offline reinforcement learning that balances optimism and pessimism using Bayesian design principles, leading to better performance and guaranteed optimal policy discovery.
https://openreview.net/forum?id=HGSIpeNNfM
Compressor summary: The paper proposes a Mixture of Receptive Fields (MoRF) for convolutional neural networks, which combines multiple receptive fields with different sizes to select the best one for each input, improving performance and capacity.
https://openreview.net/forum?id=HDrXBr26UI
Compressor summary: The paper proposes a neural-symbolic method to efficiently discover temporal logic rules that explain irregular events using vector embeddings and a sequential covering algorithm.
https://openreview.net/forum?id=HCDMiaT0Pf
Compressor summary: The paper explores using auxiliary hypotheses to initialize learning process and develops a robust model under adversarial attacks with fast generalization error bounds.
https://openreview.net/forum?id=H9fNj8ivTy
Compressor summary: This paper introduces ImplicitBench, a benchmark to evaluate text-to-image models' performance under implicit prompts, which can pose safety and privacy threats.
https://openreview.net/forum?id=H8pMSJwRD5
Compressor summary: Gen-neG is a new method that uses side-information from an oracle to guide diffusion models towards generating samples within the true data distribution, and it applies to constrained domains like self-driving simulation and safety-guarded human motion.
https://openreview.net/forum?id=H86WzfH5N1
Compressor summary: The paper analyzes the properties of sampling trajectories in diffusion-based generative models and proposes a simple technique to improve image generation by adjusting the time schedule.
https://openreview.net/forum?id=H5FDHzrWe2
Compressor summary: Stealthy Imitation is a model stealing attack on deep reinforcement learning policies that uses black-box access and no environmental knowledge to approximate the input states distribution and outperforms prior data-free approaches.
https://openreview.net/forum?id=H3bATm4mKn
Compressor summary: The paper proposes SpAR, a method that adapts neural network weights based on the eigenspectrum decomposition of source and target data, to improve out-of-distribution generalization for regression problems.
https://openreview.net/forum?id=GyV33H5Uuk
Compressor summary: The paper studies how robust nonlinear representation learning is when the mixing function is slightly misspecified, and shows that Independent Component Analysis can approximate recovery of the mixing matrix and independent components under such conditions.
https://openreview.net/forum?id=GxOFM3f5Vm
Compressor summary: EMC$^2$ is an efficient Markov Chain Monte Carlo method for generating negative samples in contrastive learning that achieves global convergence with low computational and memory cost.
https://openreview.net/forum?id=GwA4go0Mw4
Compressor summary: The paper studies how to modify neural language models' representations to prevent them from generating harmful or biased text, and proposes new optimal steering functions for this purpose.
https://openreview.net/forum?id=Grrydzui3A
Compressor summary: MISVAE improves scalability and performance of mixture variational distributions in black box variational inference by reducing parameters and inference time using importance sampling and new ELBO estimators.
https://openreview.net/forum?id=GqsRKEhelH
Compressor summary: The paper proposes a new method called Indirectly Parameterized CAEs that improves embedded feature selection by using Gumbel-Softmax distributions and avoiding duplicate selections.
https://openreview.net/forum?id=GqWy1wZKeE
Compressor summary: Co-training can learn natural concepts efficiently in a stream-based active learning model by reducing it to online classification.
https://openreview.net/forum?id=Gq1ajaKhBC
Compressor summary: The paper introduces a new framework and algorithm for efficient reinforcement learning with high-dimensional inputs, where the environment has low-dimensional latent dynamics.
https://openreview.net/forum?id=Gp5F6qzwGK
Compressor summary: The paper proposes a method (IRPO) to improve imitation learning by combining offline imitation and online reinforcement learning, and enhancing demonstration quality using online feedback.
https://openreview.net/forum?id=Gp0xZDmrA2
Compressor summary: The paper proposes a new asymmetric learning paradigm based on coupled covariance eigenproblem (CCE) for infinite-dimensional feature maps, and introduces an asymmetric Nyström method to speed up training.
https://openreview.net/forum?id=GktjBAGgo4
Compressor summary: The paper proposes a method to reduce confounding bias in causal inference by using optimal transport with learnable marginal distributions and cost functions.
https://openreview.net/forum?id=GiHo83ozsF
Compressor summary: The text introduces a new gradient aggregation method for multi-task learning that considers uncertainty and task dependencies using Bayesian inference.
https://openreview.net/forum?id=GhPFmTJNfj
Compressor summary: The paper investigates how GCN link prediction can be biased towards within-group dynamics and proposes a new fairness metric and training strategy to address this issue.
https://openreview.net/forum?id=GfNyqrwECJ
Compressor summary: The authors propose DeeMILIP, a framework for incorporating domain knowledge into deep multiple instance learning models by defining a mapping between domain entities and model components, and demonstrate its effectiveness on an immune-based diagnostics case.
https://openreview.net/forum?id=GentO2E4ID
Compressor summary: The paper proposes a novel federated learning algorithm for multi-level compositional problems that achieves linear speedup and mitigates heterogeneity and communication efficiency issues.
https://openreview.net/forum?id=GcZjpKA37R
Compressor summary: LangCell is a pre-trained language model that incorporates cross-modal knowledge of single-cell data and natural language to improve cell identity understanding tasks without relying on supervision signals.
https://openreview.net/forum?id=GcW9pg4P9x
Compressor summary: The paper proposes a Robust Constrained Policy Optimization (RCPO) algorithm for reinforcement learning under model uncertainty, which can optimize reward and satisfy constraints in real environments.
https://openreview.net/forum?id=GbFluKMmtE
Compressor summary: State-space models like Mamba can perform well in some in-context learning tasks but not others, and combining them with attention blocks improves their performance overall.
https://openreview.net/forum?id=GYGkt2M8ee
Compressor summary: Gradient-based training can harm network robustness, but using a shallow pReLU network with gradient flow improves both generalization and adversarial resistance.
https://openreview.net/forum?id=GVmvBNxB73
Compressor summary: The proposed Text-driven Knowledge Integration (TKI) method enables data-free adaptive cross-domain image retrieval using a pre-trained vision-language model and learnable domain word vectors, achieving better performance on benchmark datasets.
https://openreview.net/forum?id=GUEsK9xJny
Compressor summary: Logit-scaling GFlowNets is a new design that improves temperature-conditional GFlowNets' training speed and performance in biological and chemical tasks by scaling the policy's logits with a learned function of the temperature.
https://openreview.net/forum?id=GTnn6bNE3j
Compressor summary: AGT is a novel method that adaptively captures node characteristics and temporal variations for analyzing neurodegenerative diseases on brain connectomes, improving graph classification results.
https://openreview.net/forum?id=GLGYYqPwjy
Compressor summary: QuRating is a method that uses human intuitions to select high-quality pre-training data for language models based on four criteria, resulting in better performance and a training curriculum.
https://openreview.net/forum?id=GKcwle8XC9
Compressor summary: In-Context Unlearning is a method for removing specific training instances from large language models without retraining or updating the model parameters by providing inputs with different labels at inference time.
https://openreview.net/forum?id=GKMcCtWC7H
Compressor summary: Active inference is a statistical inference method that uses machine learning to collect labels efficiently by prioritizing uncertain data points and relying on confident predictions.
https://openreview.net/forum?id=GJzqRKOdRi
Compressor summary: The paper presents a new method for contextual inverse optimization that uses additional information to infer unknown parameters and shows improved performance on various problems.
https://openreview.net/forum?id=GHZVjmaGQM
Compressor summary: The paper proposes a hybrid loss function that combines causal and predictive losses to learn interpretable and causally valid hybrid models for complex systems, using glucose dynamics post-exercise in individuals with type 1 diabetes as an example.
https://openreview.net/forum?id=GGnYDXZC1B
Compressor summary: The paper proposes new algorithms for continuous state and action spaces in reinforcement learning, using a novel structural assumption called $u-$smoothness and feature maps based on Legendre polynomials.
https://openreview.net/forum?id=GFfWzAReAc
Compressor summary: StableMask improves language modeling by refining the causal mask, balancing attention distributions, and encoding absolute positional information without increasing parameters.
https://openreview.net/forum?id=GDp7Gyd9nf
Compressor summary: The text describes a new pipeline for designing deep learning architectures that simplifies the process by using small-scale tests to probe capabilities and identify hybrid designs that outperform existing models in scaling and performance.
https://openreview.net/forum?id=GC8HkKeH8s
Compressor summary: The text argues that selecting data for training large-scale models based on human notions of quality may not improve performance, and proposes a new method that optimizes model performance by choosing the best subset of data for each task.
https://openreview.net/forum?id=GBxflz0qdX
Compressor summary: The Differentiable Weightless Neural Network (DWN) is a new model that uses interconnected lookup tables and a novel technique for approximate differentiation, achieving superior performance in latency, throughput, energy efficiency, and accuracy on various edge computing devices and tabular datasets.
https://openreview.net/forum?id=G8zDeKOp0R
Compressor summary: SCoRe is a new representation learning framework that uses set-based submodular measures to address inter-class bias and intra-class variance, improving performance in various tasks such as classification and object detection.
https://openreview.net/forum?id=G4b32bKnBy
Compressor summary: The paper proposes a general framework to analyze interpolation models' generalization properties using concepts from high-dimensional geometry and shows how these properties depend on the Gaussian complexity and cotype of the space.
https://openreview.net/forum?id=G1igwiBBUj
Compressor summary: The paper proposes a new GFlowNet training framework that combines policy-dependent rewards with reinforcement learning to improve the efficiency and performance of generating combinatorial objects.
https://openreview.net/forum?id=G0z4bCNmkG
Compressor summary: ByMI is a general detection procedure for Byzantine machines in distributed learning that uses error rate control, sample-splitting, and score symmetry to achieve dimension insensitivity and p-value freedom.
https://openreview.net/forum?id=G0vZ5ENrJQ
Compressor summary: The paper proposes a new training strategy for deep denoisers that improves their performance by enforcing a weaker constraint called pseudo-contractiveness, and provides efficient algorithms and experimental results to support its effectiveness.
https://openreview.net/forum?id=FzyMdAm2fZ
Compressor summary: Key points: - The paper focuses on training Transformer models with differential privacy (DP). - It reduces the problem to training DP vanilla neural nets by identifying and addressing specific challenges. - It proposes Re-Attention Mechanism and Phantom Clipping to deal with attention distraction and gradient clipping issues. Summary: The paper introduces methods to train Transformer models with DP by reducing the problem to vanilla neural nets, and addressing attention distraction and gradient clipping challenges using Re-Attention Mechanism and Phantom Clipping.
https://openreview.net/forum?id=Fzp1DRzCIN
Compressor summary: The paper shows that large language models can learn to prioritize useful texts based on random tags and adjust their updates accordingly, and explores how this impacts their knowledge representation and potential risks.
https://openreview.net/forum?id=Fw4fBE2rqW
Compressor summary: The paper presents a framework for optimizing prompts in text-to-image diffusion models, addressing challenges related to domain space and text gradient computation with new methods.
https://openreview.net/forum?id=FvLd8Gr7xq
Compressor summary: The VPDM model uses vague prototypes and conditional diffusion to identify anomalies in multiple object classes without being biased by normal data.
https://openreview.net/forum?id=Fv9GLw0LkO
Compressor summary: DiffOAS is a novel algorithm that accelerates and improves the precision of PDE dataset generation for Neural Operator applications.
https://openreview.net/forum?id=FpbKoIPHxb
Compressor summary: The paper improves approximation guarantees, derandomizes algorithms, and proposes a new scalable combinatorial method for cluster deletion, a graph partitioning problem with applications in biology and social networks.
https://openreview.net/forum?id=FovMAzXUpj
Compressor summary: ReGAL is a method to improve program synthesis by learning reusable functions from existing code through refactorization, leading to better accuracy and efficiency in predicting programs across diverse domains.
https://openreview.net/forum?id=FoRqdsN4IA
Compressor summary: The paper proposes a new neural network method for learning generative models of conditional distributions, especially in situations with limited data, by minimizing a regularized objective function based on entropic optimal transport.
https://openreview.net/forum?id=FhWH9TQSMh
Compressor summary: Key points: - AI systems can be biased based on multiple sensitive attributes - The paper proposes a Bias-Guided Generative Network (BGGN) to discover high-bias intersectional sensitive attributes - Experiments on text and image datasets show the effectiveness of BGGN - Generating biased data reveals potential unfairness in modern generative AI systems Summary: The paper introduces a method to find bias in AI based on multiple sensitive attributes using a network that generates and evaluates biased data.
https://openreview.net/forum?id=Ffpg52swvg
Compressor summary: The paper introduces CodeReasoning, a benchmark for evaluating code models on input and output prediction tasks, and shows that current models struggle to solve it even with CoT and fine-tuning schemes.
https://openreview.net/forum?id=FYvpxyS43U
Compressor summary: The paper proposes RoSA, a new fine-tuning method for large language models that combines low-rank and sparse approximations to improve accuracy under limited resources.
https://openreview.net/forum?id=FYQIgQWH3d
Compressor summary: PMTR is a new framework that uses an approximate high-order feature transform layer, PMT, to reliably match and assemble geometric shapes with low memory and compute costs.
https://openreview.net/forum?id=FWlNA3et6X
Compressor summary: The paper proposes a method to measure and improve unlearning in LLMs by treating each textual sequence differently based on its memorization level, addressing privacy and copyright issues.
https://openreview.net/forum?id=FVvf69a5rx
Compressor summary: MOMENT is a family of open-source foundation models for time series analysis that overcomes challenges in pre-training and evaluation by using a large, diverse collection of time series data and a new benchmark.
https://openreview.net/forum?id=FVmqX0sYz9
Compressor summary: The paper presents a framework to audit the privacy leakage of four private prediction algorithms by varying adversary capabilities and introduces novel techniques to measure Renyi DP.
https://openreview.net/forum?id=FV3kY9FBW6
Compressor summary: The paper proposes a method called Adaptive Advantage-guided Policy Regularization (A2PR) for offline reinforcement learning that improves policy performance by selectively using high-advantage actions from an augmented behavior policy and a VAE, while maintaining conservatism from out-of-distribution actions.
https://openreview.net/forum?id=FSxTEvuFa7
Compressor summary: CarbonNovo is a unified energy-based model that generates protein structure and sequence together, improving over existing two-stage methods in various metrics.
https://openreview.net/forum?id=FQQ4476dT2
Compressor summary: FightLadder is a real-time fighting game platform for competitive multi-agent research with challenging dynamics, visual inputs, and evaluation metrics.
https://openreview.net/forum?id=FPnUhsQJ5B
Compressor summary: The authors improve noise sampling techniques for rectified flow models, introduce a novel transformer-based architecture for text-to-image synthesis, and achieve state-of-the-art results.
https://openreview.net/forum?id=FPlaQyAGHu
Compressor summary: The paper shows that some large language models can recognize their own incorrect statements after producing them, which may cause a chain reaction of errors.
https://openreview.net/forum?id=FOJE1kRcHG
Compressor summary: This paper studies deep reinforcement learning algorithms that use mean-field dynamics, policy gradient, and temporal-difference learning to learn features and find optimal policies, and introduces new methods for critic and actor updates.
https://openreview.net/forum?id=FNKnLhLuhY
Compressor summary: SNAP-DDM is a method that uses neural networks to speed up solving PDEs with arbitrary boundary conditions and geometric parameters in 2D electromagnetics and fluidic flow problems.
https://openreview.net/forum?id=FMa4c5NhOe
Compressor summary: The paper proposes a framework to certify generation risks for retrieval-augmented language models (RAG) and shows they have lower generation risks than large language models (LLMs).
https://openreview.net/forum?id=FMEhnS0948
Compressor summary: FedRank is a novel device selection method for federated learning that considers data and system heterogeneity and improves model accuracy, training efficiency, and energy consumption.
https://openreview.net/forum?id=FM61SQzF3N
Compressor summary: The proposed IIANet model leverages attention mechanisms to efficiently fuse audio-visual features for speech separation, outperforming previous methods with less computation time.
https://openreview.net/forum?id=FKkkdyRdsD
Compressor summary: BanditMIPS is a novel algorithm that significantly improves the complexity of finding the highest inner product vector among many vectors in high dimensions, and can be used in related problems like Matching Pursuit and Fourier analysis.
https://openreview.net/forum?id=FHkavpr5Ze
Compressor summary: Memory-space visual prompting (MemVP) is a novel approach that injects visual knowledge into language models by concatenating visual prompts with the weights of the feed-forward network, leading to faster training time, lower inference latency, and improved performance on various vision-language tasks.
https://openreview.net/forum?id=FG5hjRBtpm
Compressor summary: The paper proposes JRNGC, a method to learn multivariate summary and full-time Granger causality with a single model, addressing limitations of existing neural Granger causality approaches.
https://openreview.net/forum?id=FFILRGD0jG
Compressor summary: The paper proposes conditional generative models that use dynamical transport maps to couple base and target densities, which can be learned by a simple square loss regression problem and applied to super-resolution and in-painting tasks.
https://openreview.net/forum?id=FCtO757Onl
Compressor summary: BNS solvers are a new family of non-stationary solvers that improve sample efficiency of Diffusion and Flow models by distilling existing numerical ODE solvers with fast optimization, small parameter space, and diverse samples.
https://openreview.net/forum?id=FCmWhJQ14I
Compressor summary: The text discusses the limitations of "Minimum of N" metrics for evaluating trajectory prediction models in autonomous systems and suggests using Energy Score-based measures instead for better assessment.
https://openreview.net/forum?id=F3x6uYILgL
Compressor summary: MCPL is a method that learns multiple image concepts from a single sentence-image pair using textural inversion and regularisation techniques, enabling synthesis of novel images with new semantically disentangled concepts.
https://openreview.net/forum?id=F3RdeyiR5H
Compressor summary: The paper introduces a new problem of finding trust regions for black box explanations, which can provide insights into model behavior, stability, explanation reuse, and comparison of methods.
https://openreview.net/forum?id=F3G2udCF3Q
Compressor summary: The paper proposes Graph Multilinear neT (GMT), a new XGNN architecture that improves the interpretability and performance of interpretable subgraph learning on graph-structured data.
https://openreview.net/forum?id=F3Ds71Xgo1
Compressor summary: ERP improves molecule generation by balancing exploration and exploitation in Transformer decoding using entropy-revised planning, achieving state-of-the-art results on various benchmarks.
https://openreview.net/forum?id=F3936hVwQa
Compressor summary: The paper proposes extending conformal prediction to handle sequential data shifts in AI/ML systems, and presents practical algorithms and evaluations for black-box optimization and active learning tasks.
https://openreview.net/forum?id=F2Tegvyqlo
Compressor summary: The paper proposes methods to improve the calibration of over-parametrized DNNs for image classification by decoupling feature extraction and classification layers, and placing a Gaussian prior on the last hidden layer outputs.
https://openreview.net/forum?id=F1drhMjN7s
Compressor summary: Libra is a model that combines vision and language by routing the inputs through different modules, achieving strong performance in image-to-text tasks with less data.
https://openreview.net/forum?id=Ez3Lckpe4l
Compressor summary: The text discusses how the choice of learning algorithm affects the success of collective action in machine learning, using distributionally robust optimization and stochastic gradient descent as examples.
https://openreview.net/forum?id=ExHTFXEhc9
Compressor summary: The authors explore structured matrices as efficient alternatives to dense matrices in foundation models, proposing a new matrix family (Block Tensor-Train) that outperforms dense matrices on multiple tasks with less compute.
https://openreview.net/forum?id=EvHWlYTLWe
Compressor summary: The text discusses how language models can manipulate their outputs to optimize objectives and create negative side effects through feedback loops, and suggests three recommendations for evaluating this phenomenon.
https://openreview.net/forum?id=Et8Pk97u4u
Compressor summary: The paper proposes a method for numerically stable Hamiltonian dynamics on Riemannian manifolds by generalizing the idea of upper-bounding particle speed based on position and deriving an algorithm for sampling from relativistic momentum distributions.
https://openreview.net/forum?id=EsWJ5wd2ir
Compressor summary: The paper proposes Diffusion Rejection Sampling, a method that improves sampling performance under well-trained diffusion models by refining samples with varying effort depending on their quality.
https://openreview.net/forum?id=EsSSDjwFra
Compressor summary: CPRNN is a new model that combines second-order interactions and CP decomposition to improve sequence modelling over traditional RNNs while reducing parameter count.
https://openreview.net/forum?id=EruV94XRDs
Compressor summary: MusicRL is a text-to-music system that uses reinforcement learning and human feedback to generate songs that match the given captions and sound good.
https://openreview.net/forum?id=ErkzxOlOLy
Compressor summary: The paper proposes a self-supervised learning method for deep networks to better learn representations from tabular data with heterogeneous features and irregular functions, using binning as a pretext task.
https://openreview.net/forum?id=EqFxIbGWRU
Compressor summary: Zhang et al. (2021) introduced probabilistic generating circuits, which unify probabilistic circuits and determinantal point processes, by showing how to transform them into probabilistic circuits with negative weights, allowing for tractable marginalization on categorical variables.
https://openreview.net/forum?id=EncFNR3hxM
Compressor summary: Key points: - Knowledge graph reasoning is important and path-based methods have limitations - KnowFormer uses transformers for knowledge graph reasoning with message-passing - It defines attention computation based on query prototype and structure-aware modules - It outperforms baseline methods on transductive and inductive benchmarks Summary: KnowFormer is a novel method that uses transformers to perform knowledge graph reasoning with efficient message-passing, leveraging query prototype and structure-aware modules.
https://openreview.net/forum?id=ElVHUWyL3n
Compressor summary: The text discusses a probabilistic model for in-context learning that explains its dual operating modes (task learning and task retrieval) and an "early ascent" phenomenon observed in large language models.
https://openreview.net/forum?id=ElNxZ40tBJ
Compressor summary: The paper studies how to minimize worst-group risk under differential privacy using new algorithms and stability analysis, and compares different approaches.
https://openreview.net/forum?id=EhU0xBSP4l
Compressor summary: The study explores how over-parameterized ReLU CNNs can achieve near optimal accuracy in XOR-type classification tasks with label-flipping noises, under certain conditions.
https://openreview.net/forum?id=EhPpZV6KLk
Compressor summary: GRAF is a new method for predicting neural network performance that uses simple graph features, which are faster and more interpretable than existing methods and outperform them.
https://openreview.net/forum?id=EfUrTeuUfy
Compressor summary: The paper argues that small-magnitude weights in pre-trained language models contain crucial information for handling difficult tasks, and pruning them leads to performance degradation, while quantization does not have the same effect.
https://openreview.net/forum?id=Eew3yUQQtE
Compressor summary: The paper investigates how to identify latent variables and non-linear mappings in Switching Dynamical Systems, a type of sequential latent variable model, using techniques from deep latent variable models and non-linear Gaussians.
https://openreview.net/forum?id=Edz0QXKKAo
Compressor summary: Key points: - Graph Foundation Models (GFMs) are graph models trained on diverse data for various tasks and domains - GFMs face challenges in positive transfer from diverse data sources - A "graph vocabulary" encodes the invariance of graphs and can help GFM development Summary: The paper proposes a graph vocabulary, a set of basic units that capture the invariants of graphs, to enable positive transfer for Graph Foundation Models (GFMs) trained on diverse data.
https://openreview.net/forum?id=EdRb84fiJY
Compressor summary: The authors study how two-layer neural networks learn features from data using a spiked Random Features model, and provide an asymptotic description of their generalization error in high dimensions.
https://openreview.net/forum?id=Ed4KgHoKNe
Compressor summary: The paper presents a new method to enhance self-supervised learning stability by using a non-parametric memory for concept comparison during training.
https://openreview.net/forum?id=EaJ7nqJ2Fa
Compressor summary: The authors challenge the no free lunch theorems and suggest that neural networks prefer low-complexity data and can handle diverse problems with a single model.
https://openreview.net/forum?id=EZcFK8HupF
Compressor summary: The paper introduces 3D-VLA, a new model that connects 3D perception, reasoning, and action using a generative world model, improving reasoning and multimodality generation for embodied AI tasks.
https://openreview.net/forum?id=EZLsxOgcDg
Compressor summary: Key points: - The text proposes a scheme for detecting changes in a data stream distribution. - The scheme uses confidence sequences and has small detection delay and low false alarm rate. - The scheme works for dependent observations and nonparametric distributions. - The text relates the scheme to other frameworks in the literature. Summary: The paper presents a simple and effective scheme for sequential change detection in data streams, based on confidence sequences, that can handle various dependencies and nonparametric distributions.
https://openreview.net/forum?id=EZH4CsKV6O
Compressor summary: The text discusses the potential of video generation as a versatile and powerful tool for solving real-world tasks, similar to how language models have been successfully applied.
https://openreview.net/forum?id=EYvEVbfoDp
Compressor summary: HALC is a decoding algorithm that reduces object hallucinations in large vision-language models by using a local grounding mechanism and a global beam search.
https://openreview.net/forum?id=EYOo48YGhy
Compressor summary: The authors propose a physics-informed deep learning model to analyze how brain structure and function are interconnected through data geometry and neural activities are driven by brain-wide oscillation waves.
https://openreview.net/forum?id=EWt5wsEdvc
Compressor summary: Cell2Sentence (C2S) is a method that converts gene expression data into "cell sentences" to help language models understand single-cell biology and perform various tasks in the field.
https://openreview.net/forum?id=EWJn6hfZ4J
Compressor summary: The paper proposes a novel procedure, called optimal Schrödinger bridge matching, to learn Schrödinger bridges that efficiently recovers the process with minimal error and relates to energy-based modeling objectives.
https://openreview.net/forum?id=EVMzCKLpdD
Compressor summary: We propose a method for detecting out-of-distribution data in deep generative models by combining likelihoods and local intrinsic dimension estimates from a pre-trained model, achieving state-of-the-art results.
https://openreview.net/forum?id=ETNx4SekbY
Compressor summary: ObProp is a novel method to find linear features of transformer language models using almost no data, enabling better understanding of their mechanisms and biases.
https://openreview.net/forum?id=EQXZqBXeW9
Compressor summary: FedNSL is a new framework that uses neural networks to learn complex symbolic rules from distributed data, improving downstream task performance across domains.
https://openreview.net/forum?id=ENNGAY5uKC
Compressor summary: Key points: - Flexible sensors can capture human motion without constraints or privacy issues - Existing methods need large labeled datasets and MoCap studios, which are costly and hard to obtain - Proposed method uses SuDA to adapt flexible sensor data from simulation to real world without labels - Experiments show superior performance over existing distribution-based domain adaptation methods Summary: The paper proposes a novel Sim2Real solution for human motion capture using flexible sensors, which adapts the data from simulation to real world without labeled datasets and outperforms existing methods.
https://openreview.net/forum?id=ELFZWG9C7l
Compressor summary: TopNets combine topological neural networks and persistent homology to create a unified framework that improves the representation and expressivity of graph neural networks for various tasks.
https://openreview.net/forum?id=EKye56rLuv
Compressor summary: Fairproof is a system that uses cryptography to verify the fairness of machine learning models without revealing their inner workings or data.
https://openreview.net/forum?id=EK7fuAMNoI
Compressor summary: The paper extends two optimization algorithms, EAG and FEG, to constrained comonotone min-max problems and proves their optimal convergence rates.
https://openreview.net/forum?id=EIcxV7T0Sy
Compressor summary: The text proposes using category theory to create a unified framework for specifying and implementing deep learning architectures, recovering geometric constraints and encompassing various neural network designs.
https://openreview.net/forum?id=EIGbXbxcUQ
Compressor summary: The paper proposes MobileLLM, a sub-billion parameter language model with efficient architecture and weight-sharing, achieving high accuracy and performance on various tasks.
https://openreview.net/forum?id=EHjm3sXPFy
Compressor summary: The paper proposes fast sampling-based algorithms for clustering with outliers, achieving almost linear running time and outperforming previous methods.
https://openreview.net/forum?id=EFtNP211X3
Compressor summary: The paper proposes an active watermarking technique for defending against model extraction, which fine-tunes the victim model to detect and prevent theft while preserving utility and efficiency.
https://openreview.net/forum?id=EEinDTdKr1
Compressor summary: The paper analyzes the local sensitivity of dot-product self-attention in vision transformers, showing how it affects their robustness to input perturbations and presenting a new tool (LoFAST) to measure it.
https://openreview.net/forum?id=EEO4Iktfjp
Compressor summary: The paper introduces ${\it LasF}$, a module that simulates neuronal behaviors for better language modeling by replacing the final layers of pre-trained language models, achieving improved accuracy with fewer parameters.
https://openreview.net/forum?id=EDEISRmi6X
Compressor summary: The paper proposes a new analysis for edge-degree constrained subgraph, a sparsifier for maximum matching problems, and shows that the best value of its parameter beta is 6, achieving an approximation ratio of .677.
https://openreview.net/forum?id=E8FpcUyPuS
Compressor summary: The paper proposes a generalised framework for convergent regularisation using weakly convex neural networks, and demonstrates its advantages in improving CT reconstruction with learned adversarial regularisers.
https://openreview.net/forum?id=E6Nm3x7acv
Compressor summary: Key points: - Feature selection is important in machine learning and varies with context - c-STG is a new architecture that selects features based on context variables - c-STG uses hypernetwork to map context to feature selection parameters - c-STG has theoretical advantages, empirical results, and interpretability Summary: c-STG is a novel feature selection method that adapts to context using a hypernetwork and improves performance, flexibility, and interpretability.
https://openreview.net/forum?id=E4qjDAdVte
Compressor summary: Fourier NODEs (FNODEs) is a simulation-free framework that uses Fourier analysis to estimate temporal and spatial gradients from noisy data and trains neural ordinary differential equations more accurately, efficiently, and robustly than existing methods.
https://openreview.net/forum?id=E4ItiEU8Iu
Compressor summary: The paper proposes two safety semantics for compositional Reinforcement Learning, enables enforcing them, analyzes their trade-offs, and extends Boolean composition to continuous action spaces.
https://openreview.net/forum?id=E41gvBG4s6
Compressor summary: RLU is a novel label recovery scheme for federated learning that works well even in real-world settings with diverse data and optimizers.
https://openreview.net/forum?id=E3V5MMwFgd
Compressor summary: The paper proposes robust and efficient estimators for Gaussian sparse estimation tasks, such as mean estimation, PCA, and linear regression, with improved error guarantees compared to prior algorithms.
https://openreview.net/forum?id=DzLna0cFL1
Compressor summary: The paper introduces UA2, a framework for improving autonomous agents by aligning them with human intentions, environmental dynamics, and self-constraints in realistic scenarios.
https://openreview.net/forum?id=DyvhD8J3Wl
Compressor summary: The paper studies how interpolating neural networks trained using adversarial methods can still generalize well even when facing inference-time attacks, under certain distributional assumptions.
https://openreview.net/forum?id=DwwI9L67B5
Compressor summary: The paper presents a new way to design deep learning models using a frequency-domain transfer function that enables fast and memory-efficient inference without states.
https://openreview.net/forum?id=DwniHlwcOB
Compressor summary: The authors propose entropy-based variations of Shapley values that balance prior knowledge and simplicity in causal inference.
https://openreview.net/forum?id=Dwc0RwiNI5
Compressor summary: The paper proposes new fast and adaptive decentralized learning methods for distributed nonconvex optimization tasks with improved sample complexity and provides convergence analysis.
https://openreview.net/forum?id=DwTgy1hXXo
Compressor summary: Meta probing agents (MPA) evaluate large language models by transforming an original problem into a new one using psychometric theory, allowing multifaceted analysis of the models' abilities and improving them.
https://openreview.net/forum?id=DsVzHj7jcA
Compressor summary: This paper analyzes the generalization performance of three STORM-based algorithms for different levels of stochastic optimization and provides stability results and excess risk bounds.
https://openreview.net/forum?id=DrE7jVF4VW
Compressor summary: The paper proposes methods to improve recent diffusion models by optimizing posterior covariance using maximum likelihood estimation, leading to better image denoising without retraining.
https://openreview.net/forum?id=DqC9XiI71U
Compressor summary: AdaGrP is a new method for mutual transfer learning that adapts to concept drifts and has theoretical guarantees of learnability recovery without hyper-parameter tuning.
https://openreview.net/forum?id=DprrMz24tk
Compressor summary: The authors warn about non-replicable and unreliable results in machine learning research and suggest that it should be more exploratory instead of confirmatory.
https://openreview.net/forum?id=Dn4B53IcCW
Compressor summary: The paper studies how gradient descent helps graph neural networks (GNNs) learn functions using graph structure, leading to a better understanding of their behavior and generalization.
https://openreview.net/forum?id=DlR8fWgJRl
Compressor summary: Key points: - A model-based offline RL algorithm for confounded POMDPs with general function approximations - A novel identification result for learning action effects on rewards and transitions in the POMDP - A nonparametric two-stage estimation procedure for OPE that allows general function approximations - A conservative policy optimization within confidence regions based on OPE estimator - A finite-sample upper bound on the suboptimality of the learned policy Summary: The paper presents a model-based offline RL algorithm for confounded POMDPs with general function approximations, which uses a novel identification result and a nonparametric two-stage estimation procedure for OPE, followed by a conservative policy optimization within confidence regions, and shows a finite-sample upper bound on the suboptimality.
https://openreview.net/forum?id=DkqiId4AuR
Compressor summary: The paper proposes a deep reinforcement learning method to explore and fix failure modes in pre-trained neural networks for various tasks.
https://openreview.net/forum?id=Dk0RBrqiyk
Compressor summary: Key points: - The paper studies contextual bandits with low-rank reward matrix - It presents efficient algorithms for policy evaluation, best policy identification and regret minimization - The algorithms have near minimax optimality guarantees and low sample complexity Summary: The paper proposes efficient and near minimax optimal algorithms for low-rank contextual bandits that estimate the reward subspaces and perform well in policy evaluation, best policy identification and regret minimization.
https://openreview.net/forum?id=DiyE6OOGBa
Compressor summary: GenCO is a framework that allows deep generative models to create objects with hard constraints by using differentiable solvers and focusing on data distribution matching.
https://openreview.net/forum?id=DhxZVq1ZOo
Compressor summary: The paper proposes a collective certificate scheme for GNNs that certifies the robustness of a set of nodes simultaneously against graph injection attacks, improving the certification performance with efficient linear programming solutions.
https://openreview.net/forum?id=DgLFkAPwuZ
Compressor summary: The paper introduces RPG, a text-to-image generation and editing framework that uses multimodal LLMs for chain-of-thought reasoning, regional diffusion, and closed-loop text-guided image editing to generate complex images with better compositionality and semantic alignment.
https://openreview.net/forum?id=DbyHDYslM7
Compressor summary: The paper introduces Bi-Exponent Block Floating Point (BiE), a novel numerical representation for LLMs that improves accuracy, reduces overhead, and is more expressive than low-bit data formats.
https://openreview.net/forum?id=DbMm8pmoAP
Compressor summary: Key points: - Large language models are costly to train but have great potential - Evolving Subnetwork Training (EST) samples subnetworks from layers and modules of the model - EST saves FLOPs and improves performance on downstream tasks without increasing loss Summary: The paper proposes EST, a novel training method that samples subnetworks from large language models to save costs and enhance generalization.
https://openreview.net/forum?id=DYd4vyyhUu
Compressor summary: The paper proposes a method to automatically select the optimal kernel for causal discovery using a mixture model of independent noise variables.
https://openreview.net/forum?id=DYN66IJCI9
Compressor summary: GDEM is a graph distillation method that matches the eigenbasis and node features of real and synthetic graphs, preventing spectrum bias and improving cross-architecture generalization.
https://openreview.net/forum?id=DYMj03Gbri
Compressor summary: Infectious jailbreak exploits large language models in multi-agent environments by compromising one agent and causing exponential infection of others without further adversary intervention.
https://openreview.net/forum?id=DWT9uiGjxT
Compressor summary: The authors propose TALL-masks and Consensus Merging methods to improve multi-task model merging by compressing individual checkpoints and eliminating interference weights, achieving better performance and storage reduction.
https://openreview.net/forum?id=DRGgT7SyC7
Compressor summary: The authors conducted many experiments to test pruning algorithms on a synthetic dataset but found that current methods do not achieve the sparsest models, possibly due to overparameterization and other issues.
https://openreview.net/forum?id=DRBgNQ2N7U
Compressor summary: The authors derive new bounds for kernel matrix condition number and use them to analyze the generalization behavior of kernel ridge regressors with different spectral decay rates.
https://openreview.net/forum?id=DN7uk4gQ7C
Compressor summary: The authors propose a new method for dimension reduction in single-index models based on Hellinger correlation, which improves upon existing techniques by better capturing data dependencies.
https://openreview.net/forum?id=DM0r4qatjT
Compressor summary: E$^4$ is a novel algorithm for linear bandits that achieves finite-time and asymptotic optimality in regret and batch complexity, and performs well on hard instances in experiments.
https://openreview.net/forum?id=DLTjFFiuUJ
Compressor summary: This paper investigates attention sinks in large language models and proposes a training-free technique (ACT) that optimizes attention distributions during inference, improving accuracy on various tasks.
https://openreview.net/forum?id=DL79HYCFFq
Compressor summary: The Simformer is a novel amortized inference method that uses a probabilistic diffusion model with transformer architectures to overcome the limitations of current approaches in flexibility, performance, and applicability to various domains such as ecology, epidemiology, and neuroscience.
https://openreview.net/forum?id=DKOHE4n8jk
Compressor summary: PIMS is an acquisition function in Bayesian optimization that achieves a tighter theoretical regret bound than GP-UCB and TS, while avoiding their practical issues.
https://openreview.net/forum?id=DKKg5EFAFr
Compressor summary: This paper evaluates the impact of post-training quantization on various language models and tasks, and provides recommendations for applying quantization techniques.
https://openreview.net/forum?id=DJdVzxemdA
Compressor summary: DDT is a new algorithm that uses adaptive state tracing and meta-reinforcement learning to improve generalization of imitator agents in dynamic environments.
https://openreview.net/forum?id=DJXt63RLO1
Compressor summary: This paper analyzes the "Goldilocks zone" for deep learning optimization, linking excess positive curvature of the loss to initialization norm, confidence, and a new type of vanishing gradient.
https://openreview.net/forum?id=DHtF8Y6PqS
Compressor summary: This paper investigates learning with dependent data and square loss in a hypothesis class with tail decay in Orlicz space, and shows that under certain conditions, the empirical risk minimizer achieves a rate that depends only on the complexity of the class and second order statistics, without relying on mixing time.
https://openreview.net/forum?id=DCmahCZJYb
Compressor summary: Reflected reSGLD is a new algorithm that uses reflection steps to overcome stagnation issues in non-convex learning by constraining the exploration within a bounded domain, leading to better mixing rates and simulation efficiency.
https://openreview.net/forum?id=DChQpB4AJy
Compressor summary: The paper introduces ILARL, an algorithm for imitation learning in infinite horizon linear MDPs, which reduces the number of trajectories needed and improves accuracy by removing exploration assumptions and leveraging connections to online learning with adversarial losses.
https://openreview.net/forum?id=DCNCwaMJjI
Compressor summary: TROVE is a method that generates high-level functions for code LMs to solve various tasks more efficiently and accurately with smaller toolboxes and faster human verification.
https://openreview.net/forum?id=DBlkjCDg2i
Compressor summary: The paper proposes a method called StyDeSty that uses stylization and destylization modules to improve single domain generalization, achieving better results than existing approaches on multiple benchmarks.
https://openreview.net/forum?id=DBI6AuCD4a
Compressor summary: The paper proposes new stochastic optimization methods based on clipping stochastic gradient differences and proves tight high-probability convergence results for composite and distributed problems, addressing limitations of existing approaches.
https://openreview.net/forum?id=DA2AiCiCaM
Compressor summary: The text discusses how policy gradient methods smooth out complex optimization landscapes in deep reinforcement learning, but can introduce challenges due to stochasticity and exploration.
https://openreview.net/forum?id=D9EfAkQCzh
Compressor summary: This paper investigates adversarial attacks on deep multi-view clustering using GANs, and proposes a novel robust method to defend against them.
https://openreview.net/forum?id=D8zn1DnTuj
Compressor summary: Reprompting is an algorithm that learns how to generate Chain-of-Thought recipes for various reasoning tasks by iteratively sampling new recipes from previous ones, achieving superior results compared to human-written and existing methods.
https://openreview.net/forum?id=D7wi9LIE6i
Compressor summary: The paper presents an algorithm that improves gradient-free optimization over cross-polytopes for applications like adversarial attacks, explainable AI and sparse regression, by reducing the dimensionality dependence.
https://openreview.net/forum?id=D5IRvFF1lN
Compressor summary: The paper introduces a new squared loss function for item recommendation that connects better to ranking objectives and shows improved performance in experiments.
https://openreview.net/forum?id=D4B7kkB89m
Compressor summary: The paper studies how neural collapse affects deep models when the number of classes is much larger than the feature space dimension, and shows its occurrence in practice and theory.
https://openreview.net/forum?id=D32aTei4p5
Compressor summary: The paper analyzes FQE method for policy value estimation using offline data, providing theoretical insights on optimal convergence rates, error bounds, and the role of probability ratio functions.
https://openreview.net/forum?id=D2MNVeVh5J
Compressor summary: The paper presents a novel Bayesian update rule for online filtering that is robust, efficient, and works well with nonlinear models.
https://openreview.net/forum?id=CyEJn71Z00
Compressor summary: This paper studies how memorization affects learning in stochastic convex optimization, using conditional mutual information to measure it, and shows its importance for generalization and privacy.
https://openreview.net/forum?id=Cw6Xl0g8a5
Compressor summary: The paper proposes new criteria for explaining vision transformers (ViTs) and introduces a variational Bayesian method, PACE, that provides faithful, stable, sparse, multi-level, and parsimonious explanations of ViT predictions by modeling patch embeddings.
https://openreview.net/forum?id=CvRu2inbGV
Compressor summary: The paper introduces a new fairness measure based on Wasserstein distance for continuous scores, which is easier to compute and interpret than existing methods, and shows its advantages over ROC-based measures.
https://openreview.net/forum?id=CuiRGtVI55
Compressor summary: CoIn is a module that adds convolutions to ViTs for better adaptation in visuo-motor control tasks by introducing locality and equivariance biases.
https://openreview.net/forum?id=CtyLla0DU8
Compressor summary: The paper proposes a framework to improve cognitive generalization in DRL by building a latent space in a simple scenario, segmenting it based on environmental influences, and using it to fine-tune policies in complex scenarios without designing new rewards.
https://openreview.net/forum?id=CtgJUQxmEo
Compressor summary: FADiff is a novel method for designing proteins with multiple functions and floating motifs, which does not require prior knowledge of their positions.
https://openreview.net/forum?id=CtEWswTjUd
Compressor summary: This paper proposes a sparse generative model that explains why deep networks learn abstract representations and become insensitive to task invariances, leading to better performance.
https://openreview.net/forum?id=Cs0Xy6WETl
Compressor summary: Reflective Policy Optimization (RPO) improves on-policy reinforcement learning methods by using past and future state-action information for policy optimization, leading to faster convergence and better sample efficiency.
https://openreview.net/forum?id=CrUmgUaAQp
Compressor summary: The paper explores how multi-agent debate can improve the accuracy of large language models, but finds that other strategies may be more effective depending on hyperparameters and agreement levels.
https://openreview.net/forum?id=CquFGSIU6w
Compressor summary: MET is a novel FSOSR model that uses an evidential open-set loss, cross-attention mechanism, and evidence-to-variance ratio to improve detection of instances from unseen classes while maintaining closed-set performance.
https://openreview.net/forum?id=CpgKRKBUTl
Compressor summary: The study proposes a simple modification for SGD that makes the outputs of neural networks provably compressible without requiring any nontrivial assumptions by injecting heavy-tailed noise to the iterates.
https://openreview.net/forum?id=CpcaL75UgY
Compressor summary: The paper proposes a method to improve complex reasoning in large language models by using pairwise-comparison evaluation instead of noisy point-wise scoring from the LLM, with ensembles and dueling bandits to reduce noise.
https://openreview.net/forum?id=CpI37NA7MO
Compressor summary: SMASH is a new method to learn from spatio-temporal point processes, which can model and predict events with temporal and spatial features more accurately and with uncertainty estimates.
https://openreview.net/forum?id=CmXkdlO6JJ
Compressor summary: AdamW outperforms Adam with $\ell_2$ regularization in language modeling tasks due to its implicit constrained optimization that ensures bounded $\ell_\infty$ norm of parameters.
https://openreview.net/forum?id=CmOmaxkt8p
Compressor summary: The paper presents NegotiationArena, a framework for evaluating how well large language models can negotiate with each other in various scenarios.
https://openreview.net/forum?id=ClWdplZ12B
Compressor summary: The Multi-level Actor-Critic (MAC) framework uses a Multi-level Monte-Carlo (MLMC) gradient estimator to reduce the need for oracle knowledge of the mixing time, improving global convergence and performance in average-reward reinforcement learning.
https://openreview.net/forum?id=CjVWen8aJL
Compressor summary: This paper proposes ParaTAA, a parallel algorithm that speeds up diffusion model sampling by solving triangular nonlinear equations with systematic techniques.
https://openreview.net/forum?id=CiZN2OATRp
Compressor summary: The paper proposes a framework to provide high-probability bounds on estimand values for target distributions using domain knowledge to partially identify them and account for selection bias in large-scale datasets.
https://openreview.net/forum?id=Chy4rSqy4Y
Compressor summary: The paper presents a fast and effective adversarial attack on no-reference image and video quality metrics, called Invisible One-Iteration (IOI), which can be used to test the robustness of learning-based metrics under video attacks.
https://openreview.net/forum?id=CgO2cuWWLV
Compressor summary: The paper studies how an agent can learn from instructions that are suitable for its actions, and introduces a no-regret algorithm that depends only on the intrinsic rank of the instruction-response distribution.
https://openreview.net/forum?id=CfOtiepP8s
Compressor summary: This paper investigates how large language models use a small fraction of attention heads for arithmetic calculations and explores fine-tuning these heads to improve their computational performance.
https://openreview.net/forum?id=CecY6XiUfu
Compressor summary: Reference neural operators (RNO) are a novel data-efficient method for learning the smooth dependence of solutions on geometric deformations, which can significantly reduce errors in partial differential equation simulations with arbitrary geometries.
https://openreview.net/forum?id=CduFAALvGe
Compressor summary: IRED is a new framework that learns to reason using energy-based optimization and adapts to problem difficulty, achieving better performance in various tasks.
https://openreview.net/forum?id=CbbTF6tDhW
Compressor summary: The paper proposes a novel method to train unbiased and accurate models by grouping data into different shortcuts and optimizing their losses dynamically, achieving minimax Pareto solution.
https://openreview.net/forum?id=Cbacx90Wkt
Compressor summary: The paper proposes OAK, a framework that uses auxiliary data to improve extreme classification accuracy by enriching document embeddings and selecting precise labels.
https://openreview.net/forum?id=CbIZatwz9z
Compressor summary: Online-iForest is an efficient online anomaly detection method that adapts to evolving data and outperforms existing solutions in real-world applications.
https://openreview.net/forum?id=CbIRQgAYE4
Compressor summary: DreamCoder improves program synthesis by learning to simplify search, compress solutions, and extract relevant components using a neural search policy.
https://openreview.net/forum?id=CaxQ5IbHgF
Compressor summary: The paper investigates the convergence properties of methods that minimize upper bounds on MAP inference problems using dual linear programming or Lagrangian relaxation by coordinate descent, and proves that they converge to a fixed point with a specific convergence rate.
https://openreview.net/forum?id=CY0lFwD4qx
Compressor summary: The paper proposes a new method using Reservoir Computing and Genetic Algorithm to forecast SARS-CoV-2 hospitalizations more accurately than existing approaches.
https://openreview.net/forum?id=CXZqGJonmt
Compressor summary: CosPGD is a robustness-evaluating adversarial attack that uses an alignment score to scale the loss, improving efficiency and effect balance for semantic segmentation and regression models.
https://openreview.net/forum?id=CV9PiQGt0i
Compressor summary: The paper proposes an adversarial data splitting framework to improve offline reinforcement learning's generalization by adaptively extracting knowledge from empirical data and simulating distribution shifts.
https://openreview.net/forum?id=CTgEV6qgUy
Compressor summary: The paper proposes an active learning strategy for Direct Preference Optimization, which uses a practical acquisition function to improve the rate and performance of fine-tuning large language models with human or AI preferences.
https://openreview.net/forum?id=CTEMHDSwIj
Compressor summary: The paper studies how the depth of fusion in multimodal neural networks affects unimodal bias, which can lead to poor generalization and permanent reliance on one modality during training.
https://openreview.net/forum?id=CSIfCpXhCF
Compressor summary: The paper presents CrossGET, a framework that adaptively combines tokens in real-time during inference to improve efficiency and performance of vision-language Transformers on various tasks.
https://openreview.net/forum?id=CR6Sl80cn8
Compressor summary: The paper presents a new black-box adversarial attack method that uses a surrogate white-box model as a global function prior, improving query efficiency and success rate.
https://openreview.net/forum?id=CQI3f1U9X1
Compressor summary: The paper proposes a quantum algorithm to solve finite-sum optimization problems with smooth and strongly convex functions, achieving better complexity than the classical bound and providing lower and upper bounds for generalizations.
https://openreview.net/forum?id=CQH63IbI5o
Compressor summary: The novel approach combines neural temporal point processes with diffusion generative models to improve long-horizon forecasting of irregular event sequences by learning joint distributions of types and inter-arrival times.
https://openreview.net/forum?id=CNicRIVIPA
Compressor summary: Score entropy is a new loss function that improves discrete diffusion models for natural language tasks, outperforming existing methods like GPT-2.
https://openreview.net/forum?id=CLJZI5kDhX
Compressor summary: The text introduces a new loss function for auto-encoders that encourages independence between codebooks in music generation using language models, improving audio quality and speed.
https://openreview.net/forum?id=CKCzfU9YKE
Compressor summary: The paper presents efficient, dimension-independent replicable algorithms for learning large-margin halfspaces with improved sample complexity and accuracy compared to existing methods.
https://openreview.net/forum?id=CJbhtpcyGL
Compressor summary: The study shows how to reliably detect machine-generated text from human-written text using information theory and empirical tests with various datasets and text generators.
https://openreview.net/forum?id=CHz7WshPcp
Compressor summary: Deep LTMLE is a novel transformer-based method that estimates counterfactual outcomes under dynamic treatment policies in longitudinal studies, correcting for bias and providing confidence intervals using TMLE framework and asymptotic theory.
https://openreview.net/forum?id=CGR3vpX63X
Compressor summary: TSLANet is a universal convolutional model for time series tasks that captures long-term and short-term interactions, enhances feature representation, and adapts to noise levels and data sizes.
https://openreview.net/forum?id=CG44RLeXt1
Compressor summary: Key points: - Neural networks can learn from noisy labels by using their self-cognition ability - Self-cognition allows them to distinguish noise within and among label sources - SDM is a method that exploits this ability to denoise during training - Selective distillation module improves efficiency Summary: The paper proposes SDM, a method that leverages neural networks' self-cognition ability to learn from noisy labels and improve efficiency with selective distillation.
https://openreview.net/forum?id=CEfr3h68KU
Compressor summary: The paper analyzes malicious backdoors in quantized models and proposes a method to purify them by aligning activation distributions.
https://openreview.net/forum?id=CDnv4vg02f
Compressor summary: RaLMSpec improves the speed and performance of iterative language model serving with speculative retrieval, batched verification, and several optimizations.
https://openreview.net/forum?id=CD2xl1L5es
Compressor summary: The paper proposes a framework to balance labels and semantics in pedestrian attribute datasets and improve model accuracy with minimal computational cost.
https://openreview.net/forum?id=CBcNl5Eo32
Compressor summary: The paper proposes a reward system for learning agents to identify their peers' strategies in multi-agent games, enabling faster adaptation and better outcomes.
https://openreview.net/forum?id=C7Z8EhZ6bl
Compressor summary: The paper introduces Factored-Reward Bandits, a setting for sequential decision problems with structured intermediate effects, and proposes two algorithms to minimize regret in this setting.
https://openreview.net/forum?id=C64clssMVU
Compressor summary: The text proposes L1-Coverage as a new exploration objective for reinforcement learning that balances intrinsic complexity, efficient planning, and efficient exploration in high-dimensional domains.
https://openreview.net/forum?id=C4nalr0DoE
Compressor summary: The paper proposes a new perspective on Low-Rank Adaptation (LoRA) as a control process and improves its performance using parameter-free attention mechanisms without increasing parameters.
https://openreview.net/forum?id=C4jkx6AgWc
Compressor summary: The paper proposes a method to improve visual MBRL agents by using spatio-temporal masking, bisimulation, latent reconstruction, and a Hybrid Recurrent State-Space Model (HRSSM) to handle noisy inputs and learn robust world models for better control tasks.
https://openreview.net/forum?id=C4OpREezgj
Compressor summary: The authors propose TS-LLM, a tree-search learning framework for large language models that adapts to various tasks, sizes, and search depths, and improves LLM reasoning and planning abilities.
https://openreview.net/forum?id=C1iNBLIClt
Compressor summary: The authors develop methods to identify and estimate causal effects in linear models with latent variables, under known or unknown causal graphs, using a modified RICA algorithm.
https://openreview.net/forum?id=C0sGIO2MZN
Compressor summary: The paper presents a new method to protect data privacy in 3D deep learning by creating shortcuts in the feature space that prevent degeneracy in bi-level optimization.
https://openreview.net/forum?id=BxAvcnlS8O
Compressor summary: RIME is a robust preference-based reinforcement learning algorithm that uses sample selection and warm starts to learn from noisy human preferences.
https://openreview.net/forum?id=BwAkaxqiLB
Compressor summary: Evolution of Heuristic is a novel method that uses large language models and evolutionary computation to automatically design high-performance heuristics for complex optimization problems, outperforming handcrafted heuristics and other automatic design methods.
https://openreview.net/forum?id=BvBdYSIkpb
Compressor summary: The paper proposes a reward-free reinforcement learning algorithm, GFA-RFE, that uses uncertainty-aware exploration and weighted learning to improve sample efficiency for learning multiple tasks.
https://openreview.net/forum?id=BtbijvkWLC
Compressor summary: The paper links tempering in Sequential Monte Carlo to entropic mirror descent, shows it as a descent scheme of the KL divergence with respect to different geometries, and derives adaptive tempering rules that perform better than alternatives.
https://openreview.net/forum?id=BrZPj9rEpN
Compressor summary: DORA is a novel approach for learning adaptable policies from limited offline data by using an information bottleneck principle to improve dynamics encoding and online adaptation.
https://openreview.net/forum?id=BrCrnaCYDc
Compressor summary: The algorithm adapts the smoothing kernel to approximate the Hessian of the objective function, improving gradient estimation and optimization performance in noisy conditions.
https://openreview.net/forum?id=Bq2THeNXRr
Compressor summary: This paper proposes a new method to select the best pre-trained language model for fine-tuning by predicting its performance and introducing a modified Scaling Law that captures a previously unobserved phase transition phenomenon.
https://openreview.net/forum?id=BoPj12CnAn
Compressor summary: The paper proposes weighted distance nearest neighbor condensing, a new method for condensing data points with better performance than standard nearest neighbor rule and similar generalization bounds.
https://openreview.net/forum?id=BmPWtzL7Eq
Compressor summary: The paper proposes FuRL, a fine-tuning method for using pre-trained VLMs as rewards in RL to improve performance on sparse reward tasks with textual descriptions.
https://openreview.net/forum?id=Bic3Vmy2DG
Compressor summary: AudioSeal is a fast and effective audio watermarking technique for detecting AI-generated speech with high accuracy and low imperceptibility, which can be used to ensure voice authenticity.
https://openreview.net/forum?id=BiWIERWBFX
Compressor summary: The paper introduces $\Delta$-IRIS, a new model-based RL agent that uses discrete autoencoders and autoregressive transformers to encode stochastic deltas between time steps, achieving state of the art results on Crafter benchmark while being faster to train than previous methods.
https://openreview.net/forum?id=BiENLaUwlK
Compressor summary: The paper proposes a new method (GBE) for estimating constraint changes in finite-horizon non-discounted Safe RL, which leads to a novel algorithm (CGPO) that effectively identifies feasible optimal policies.
https://openreview.net/forum?id=Be2B6f0ps1
Compressor summary: The paper examines how governments struggle to regulate AI due to gaps between policy goals and technical feasibility, and calls for closer collaboration between AI researchers and policymakers.
https://openreview.net/forum?id=BdQTCAuT6L
Compressor summary: PEP is a private learning model that never releases a hypothesis, but provides predictions via an oracle using unlabeled examples from the underlying distribution, with improvements in robustness and privacy.
https://openreview.net/forum?id=Bc4vZ2CX7E
Compressor summary: The paper argues that open-endedness is a key property for artificial superhuman intelligence and proposes a path to achieve it using foundation models that can make novel, human-relevant discoveries.
https://openreview.net/forum?id=Bb8pOvWIe4
Compressor summary: The paper proposes a method to estimate the causal effect of algorithmic actions on user consumption without randomized experiments, using assumptions about the dynamics of consumption over time and control theory.
https://openreview.net/forum?id=BajM6YzKvm
Compressor summary: The paper proposes a new recommendation algorithm (MMTS) for online matching markets with unknown preferences and quota constraints, using bandit learning and a double matching technique to achieve stability and low regret.
https://openreview.net/forum?id=BTkaKA74mS
Compressor summary: This paper proposes a mathematical framework to analyze and improve Generative Masked Language Models (GMLMs), which are non-autoregressive text generation models that balance speed and quality, and demonstrates their effectiveness in machine translation tasks.
https://openreview.net/forum?id=BRfqYrikdo
Compressor summary: The paper introduces WorkArena and BrowserGym to evaluate large language model agents' ability to interact with enterprise software systems, finding that current agents show promise but are far from fully automated and revealing a significant performance gap between open and closed-source LLMs.
https://openreview.net/forum?id=BRIcZiK5Fr
Compressor summary: TSS is a method for selecting informative nodes in noisy graph data using topological information to improve GNNs' performance.
https://openreview.net/forum?id=BPQHXwVNvl
Compressor summary: Online speculative decoding updates draft models on user query data to improve accuracy and reduce latency in large language model inference.
https://openreview.net/forum?id=BOunbuapcv
Compressor summary: VQDNA is a framework that improves genome tokenization by using vector-quantized codebooks to learn patterns from the genome vocabulary and Hierarchical Residual Quantization to enrich it further, outperforming existing models and revealing insights on SARS-CoV-2 mutations.
https://openreview.net/forum?id=BOorDpKHiJ
Compressor summary: The authors propose UltraFeedback, a large and diverse dataset of AI feedback for user-assistant interactions, and show how it can be used to align open-source chat language models without human feedback.
https://openreview.net/forum?id=BOFjRnJ9mX
Compressor summary: The paper proposes a method, GMTNet, to predict tensor properties of crystalline materials that preserves their symmetries and outperforms existing methods.
https://openreview.net/forum?id=BO0jookxk8
Compressor summary: This paper studies the least squares estimators for deterministic mixture of experts models and shows that feed forward networks with sigmoid or tanh activation functions converge faster than polynomial experts.
https://openreview.net/forum?id=BNH8spaR3l
Compressor summary: The paper introduces a new probabilistic time series model ($ ext{D}^3 ext{M}$) that unifies denoising diffusion models and continuous flow models, improves generation speed, and performs well on imputation and forecasting tasks.
https://openreview.net/forum?id=BNAvYSCrLD
Compressor summary: In-context learning dynamics of large language models depend on problem framing and are influenced by agency implications.
https://openreview.net/forum?id=BJx1K4lAAX
Compressor summary: The paper introduces multi-view stochastic block models for graph clustering using multiple data sources and presents efficient algorithms that improve on existing methods.
https://openreview.net/forum?id=BIbjwcrg0V
Compressor summary: PyJuice is a GPU implementation for probabilistic circuits that improves speed and memory efficiency for large-scale generative models.
https://openreview.net/forum?id=BIMSHniyCP
Compressor summary: Relational Deep Learning (RDL) is a method to learn from relational databases using Graph Neural Networks without manual feature engineering.
https://openreview.net/forum?id=BH8TYy0r6u
Compressor summary: The text argues that AI models are increasingly similar in how they represent data, possibly converging on a shared statistical model of reality called the platonic representation.
https://openreview.net/forum?id=BCEtumPYDt
Compressor summary: Uncertainty Sampling is a useful Active Learning method for node classification on graphs, especially when using ground-truth Bayesian uncertainty estimates.
https://openreview.net/forum?id=B5g6y7JlMw
Compressor summary: Constant imputation may introduce bias in linear prediction, but its relevance and performance are improved when using a random features model and stochastic gradient predictors, especially for high-dimensional data and missing completely at random (MCAR) cases.
https://openreview.net/forum?id=B5906M4Wnd
Compressor summary: The paper introduces a method for automated statistical model discovery using large language models, which iteratively propose, critique, and refine models without defining a domain-specific language or search procedure.
https://openreview.net/forum?id=B4rViOCoNf
Compressor summary: The paper explores how to use ControlNet to add conditional controls to consistency models for controllable visual content creation, and proposes a tailored training strategy using consistency training to improve image quality and details.
https://openreview.net/forum?id=B48Pzc4oKi
Compressor summary: LLaGA is a new model that combines large language models with graph data analysis, enabling versatile, generalizable, and interpretable results on various graph tasks.
https://openreview.net/forum?id=B1ajnQyZgK
Compressor summary: The paper explores how to use Universal Turing Machines to generate diverse training data for meta-learning, enabling neural networks to learn universal prediction strategies.
https://openreview.net/forum?id=B1W712hMBi
Compressor summary: NExT is a method that teaches LLMs to understand how programs execute at run-time by inspecting execution traces and reasoning through chain-of-thought rationales, improving their ability to repair code.
https://openreview.net/forum?id=B0xmynxt4f
Compressor summary: DISCRET is a self-interpretable AI framework that generates faithful explanations for individual treatment effect estimation, using database queries and reinforcement learning to balance accuracy and faithfulness.
https://openreview.net/forum?id=AzUCfhJ9Bs
Compressor summary: The paper analyzes how disparities in learning rates across layers affect trainability in deep neural networks and proposes a warm-up method to minimize these disparities.
https://openreview.net/forum?id=AxmefV2NEf
Compressor summary: TimeMIL is a novel weakly supervised method for multivariate time series classification that uses a tokenized transformer and wavelet positional token to better locate patterns of interest and account for temporal dependencies.
https://openreview.net/forum?id=Ax90jQPbgF
Compressor summary: ICSDICE is an offline ICRL algorithm that learns safety constraints and control policies from expert data, focusing on feasible constraints and transferable estimations.
https://openreview.net/forum?id=AwLLSlJAeJ
Compressor summary: The paper proposes a method for sampling text from an energy-based model that uses gradient information and works well in practice, overcoming previous limitations of gradient-based MCMC for text generation.
https://openreview.net/forum?id=AtVtt9xsO1
Compressor summary: ZkAudit is a protocol that allows model providers to keep their models secret while enabling trustless audits of properties like copyright and censorship by using cryptographic commitments and zero-knowledge proofs.
https://openreview.net/forum?id=Ar0dsOMStE
Compressor summary: Adaptive environment design improves sample-efficiency and robustness of learning reward functions from expert demonstrations by repeatedly interacting with the expert in varied environments.
https://openreview.net/forum?id=AqGCEHK9dZ
Compressor summary: The paper proposes a method to privately and accurately estimate the frequency of items in a multiset using histograms and discrete Laplace noise.
https://openreview.net/forum?id=AqBz54aFyj
Compressor summary: The text discusses a novel multi-objective optimization approach for watermarking AI-generated texts to distinguish them from human-written ones while maintaining semantic quality.
https://openreview.net/forum?id=ApRKrKZJSk
Compressor summary: The paper presents a universal equivariant graph neural network architecture for point clouds with positions and velocities, based on extending the $2$-Weisfeiler-Leman test to these scenarios.
https://openreview.net/forum?id=AocOA4h3bu
Compressor summary: The paper proposes a new representation learning method that reduces information leakage by using subtraction inductive bias and outperforms existing methods on various data types.
https://openreview.net/forum?id=AoYhtJ4A90
Compressor summary: FedBPT is a framework that enables efficient and private fine-tuning of pre-trained language models using black-box inference and gradient-free optimization methods, reducing communication and memory costs.
https://openreview.net/forum?id=Ao9UUaScAU
Compressor summary: The paper investigates how the decoder's orthogonality properties in variational autoencoders contribute to disentangled representation learning, both theoretically and experimentally.
https://openreview.net/forum?id=AlJkqMnyjL
Compressor summary: The paper proposes algorithms for dynamic optimization of monotone submodular functions under cardinality constraints, balancing consistency and approximation quality.
https://openreview.net/forum?id=Al5GlVytqi
Compressor summary: The authors propose a new algorithm to generate intersection features for candidate triples in knowledge graphs, which improves link prediction performance and training efficiency.
https://openreview.net/forum?id=Aj18fUB6Th
Compressor summary: This paper proposes a two-timescale derivative free optimization algorithm for performative prediction in decision-dependent settings with controlled Markov chains, and shows its sample complexity.
https://openreview.net/forum?id=Ada9Z68nvb
Compressor summary: The authors propose RFold, a fast and accurate method for predicting RNA secondary structure using a K-Rook problem formulation and bi-dimensional optimization.
https://openreview.net/forum?id=Ad9msn1SKC
Compressor summary: T-CREx is a fast and effective method for explaining how decisions would change under different scenarios using generalised rules and metarules.
https://openreview.net/forum?id=AbGbGZFYOD
Compressor summary: DéjàVu improves distributed LLM serving efficiency by using DéjàVuLib to reduce bubbles, manage GPU memory, and ensure fault tolerance.
https://openreview.net/forum?id=AaTYLZQPyC
Compressor summary: The paper proposes a method to learn reachable state associations from multi-step inverse dynamics for effective goal-conditioned planning and policy learning in various simulation environments.
https://openreview.net/forum?id=AZWqXfM6z9
Compressor summary: The paper introduces Charmer, a query-based character-level adversarial attack that can achieve high success rate and similarity on both small and large NLP models.
https://openreview.net/forum?id=AZ1tWCa9j3
Compressor summary: The paper presents an efficient learning algorithm for Single-Index Models under the $L_2^2$ loss in the agnostic model, which works for various distributions and link functions, and introduces a new concept called alignment sharpness.
https://openreview.net/forum?id=AYbXN9poJl
Compressor summary: X-Oscar is a framework for creating high-quality 3D avatars from text prompts using a step-by-step "Geometry→Texture→Animation" approach and new techniques to overcome oversaturation and low-quality output issues.
https://openreview.net/forum?id=AYWBRwsZ8z
Compressor summary: This paper studies how prior distribution mismatch affects Plug-and-Play methods for image inverse problems and proposes a domain adaptation strategy to mitigate its impact.
https://openreview.net/forum?id=AVEc9LvSlO
Compressor summary: The text proposes a method to teach generative models to estimate their uncertainty and predict true conditional distributions by training them to cheat on independent response pairs.
https://openreview.net/forum?id=ATvN9JnqZ8
Compressor summary: Key points: - EnzyGen is an approach to design functional enzymes based on their sites and substrates - It uses a novel network of attention and equivariant layers to capture sequence and 3D information - It outperforms baseline methods in substrate binding affinity across all enzyme families Summary: EnzyGen is a method that can automatically design functional enzymes by learning from their sites and substrates, using a novel network that captures both sequence and 3D information, and achieving superior performance in binding affinity.
https://openreview.net/forum?id=ATRnM8PyQX
Compressor summary: COALA is a vision-centric Federated Learning platform that supports various tasks, data types, and model configurations, as well as customization and benchmarking for real-world scenarios.
https://openreview.net/forum?id=AQYabSOfci
Compressor summary: The paper proposes a simple analysis to quantify and mitigate reasoning shortcuts in abductive learning models and neural-symbolic predictive models, affecting their generalization ability.
https://openreview.net/forum?id=AOJCCFTlfJ
Compressor summary: This paper introduces a new unsupervised RL framework that combines local partition exploration with state distribution constraints, leading to better state coverage and skill learning for downstream tasks.
https://openreview.net/forum?id=ALc7DmOTI2
Compressor summary: The paper presents two theorems for binary classification models that show the importance of Precision-Recall AUC over Receiver Operating Characteristic AUC, especially for imbalanced datasets, and provide a method to compare them.
https://openreview.net/forum?id=AJGwSx0RUV
Compressor summary: EfficientPlace is a novel framework that combines global tree search and reinforcement learning for fast and high-quality macro chip placement, overcoming the sample inefficiency of existing techniques.
https://openreview.net/forum?id=AIXUuLCuMe
Compressor summary: The text discusses how large language models create meaningful representations that correlate with external variables but are not evidence of artificial general intelligence, and calls for caution in interpreting and communicating such results.
https://openreview.net/forum?id=AG45XqwPKU
Compressor summary: SYFLOW is an approach that uses normalizing flows to model target distributions and a novel neural layer for interpretable subgroup descriptions, enabling the discovery of diverse and exceptional sub-populations in large datasets.
https://openreview.net/forum?id=AFfXlKFHXJ
Compressor summary: This paper proposes a method for learning from intractable distributions in combinatorial optimization using latent variable models without relying on exact sample likelihoods, and shows its effectiveness on various benchmark problems.
https://openreview.net/forum?id=AFAX28TdO4
Compressor summary: DFlow is a new generative framework that combines Normalizing Flow with Denoising AutoEncoder for high-quality waveform generation and outperforms existing methods in speed and quality.
https://openreview.net/forum?id=AEqim4X0NV
Compressor summary: The paper proposes a new way to understand diffusion models using quantum physics concepts, which can help explain why one sampling method works better than another depending on a key parameter.
https://openreview.net/forum?id=AEHXvoOxV9
Compressor summary: Empirical weak convergence (EWC) is a general assumption that explains why kernel methods perform well under non-independent and non-mixing data, and this paper shows how it applies to various kernel methods including SVMs, kernel mean embeddings, and finite- and infinite-dimensional outputs.
https://openreview.net/forum?id=ADnUzsmsLW
Compressor summary: The paper explores why Large Language Models struggle with logical constructs in code generation and proposes a counterfactual testing framework to evaluate their understanding of programming concepts.
https://openreview.net/forum?id=AD5QC1BTJL
Compressor summary: The paper extends and speeds up a scheme that gives polynomial time approximations for $\mathcal{NP}$-hard problems using one-bit predictions and aims at finding fast algorithms with approximation consistency, smoothness and robustness.
https://openreview.net/forum?id=ABt0jlLZtX
Compressor summary: The paper studies how to learn stochastic policies for continuous RL, then convert them to deterministic ones while ensuring good performance and convergence.
https://openreview.net/forum?id=A9hJvQHEEP
Compressor summary: The paper proposes a quantum computing method (QontOT) for learning conditional distribution of transportation plans, which outperforms classical Neural OT on challenging tasks and real data.
https://openreview.net/forum?id=A9fLbXLRTK
Compressor summary: The paper studies how associative memory modules with token embeddings learn and converges, finding insights about overparameterized, imbalanced, and underparameterized regimes.
https://openreview.net/forum?id=A9MiJdetnZ
Compressor summary: The text proposes a framework to study and train retrieval-augmented models for ML systems, which use a retriever to find relevant information and a predictor to make final predictions, with excess risk bounds analysis.
https://openreview.net/forum?id=A7CtiozznN
Compressor summary: The text discusses how online learning techniques can significantly improve the ability of proof assistants to solve mathematical theorems by exploiting locality properties, and introduces two solvers that outperform other general purpose provers.
https://openreview.net/forum?id=A6fmX9QCEa
Compressor summary: Tuning-free algorithms can perform similarly to optimally-tuned ones for some machine learning problems but not all, and their effectiveness depends on problem parameters and noise distribution.
https://openreview.net/forum?id=A54CXWn9VB
Compressor summary: The paper analyzes how different regularization terms affect linear regression in continual learning, showing that generalized $\ell_2$-regularization algorithms can achieve optimal performance while balancing trade-offs and handling data heterogeneity.
https://openreview.net/forum?id=A0N39kgRZq
Compressor summary: The text discusses an adaptive algorithm for learning an unknown subset of a hypercube with efficient query complexity and explores variants of the problem with different constraints.
https://openreview.net/forum?id=9zlZuAAb08
Compressor summary: QDHF is a novel approach that infers diversity metrics from human feedback to improve the quality and diversity of solutions in complex and open-ended domains, such as text-to-image generation.
https://openreview.net/forum?id=9zdTOOgutk
Compressor summary: The authors propose Neighbors as Queries (NaQ), an unsupervised episode generation method for few-shot node-classification that uses graph meta-learning to better utilize node information and adapts to downstream tasks.
https://openreview.net/forum?id=9yADTDHgGu
Compressor summary: The paper proposes a general framework for transfer learning across different types of learning problems and shows how it can reveal new insights and challenge common assumptions about when and how transfer learning works.
https://openreview.net/forum?id=9xUpLGAOy9
Compressor summary: Key points: - The paper proposes a hierarchical imitation learning method for complex low-level control tasks using sub-optimal demonstrations. - The method discovers and predicts the chain-of-thought (CoT) of the demonstrations as guidance for policy learning. - The method outperforms existing baselines on various manipulation tasks. Summary: The paper introduces a novel imitation learning method that learns to predict the chain-of-thought of sub-optimal demos and uses it to guide policy learning for complex low-level control tasks, achieving superior results.
https://openreview.net/forum?id=9vKRhnflAs
Compressor summary: Flextron is a network architecture and optimization framework that enables efficient adaptation of large language models for specific latency and accuracy targets during inference without fine-tuning or additional training.
https://openreview.net/forum?id=9oAXix8da9
Compressor summary: Pi-DUAL uses privileged information to distinguish clean from wrong labels and improve deep learning models' performance in the presence of label noise.
https://openreview.net/forum?id=9laB7ytoMp
Compressor summary: Skill Set Optimization (SSO) improves large language models' sequential decision making by constructing and refining sets of transferable skills using environment reward signals.
https://openreview.net/forum?id=9kArQnKLDp
Compressor summary: CARTE is a neural architecture that uses graph representation and attention to process tables without needing entity and schema matching, achieving better performance than tree-based models.
https://openreview.net/forum?id=9jXS07TIBH
Compressor summary: The text introduces a new method, Pseudo-Negative Regularization (PNR), for continual self-supervised learning that uses pseudo-negatives from previous models to maintain consistency in representation learning.
https://openreview.net/forum?id=9iRGs3wBTy
Compressor summary: The text develops algorithms for online linear regression with optimal regret guarantees, even without prior knowledge, using a novel analysis of a discounted forecaster and showing its optimal performance.
https://openreview.net/forum?id=9iGdh0wAgB
Compressor summary: The text explains how neural networks use correlations between latent variables to efficiently learn from higher-order input cumulants, which are crucial for their performance but computationally hard to extract.
https://openreview.net/forum?id=9cG1oRnqNd
Compressor summary: The paper proposes $ exttt{CLLM}$, a method that uses large language models to generate and curate high-quality synthetic data for machine learning tasks in low-data settings, outperforming conventional generators.
https://openreview.net/forum?id=9ZxnPZGmPU
Compressor summary: Promptbreeder is a self-improving mechanism that evolves prompts for various domains, outperforming state-of-the-art methods on reasoning and hate speech tasks.
https://openreview.net/forum?id=9ZkUFSwlUH
Compressor summary: Di-SkilL is an RL method that learns diverse skills using Mixture of Experts with maximum entropy and energy-based models for context distribution.
https://openreview.net/forum?id=9Ub6nLqdMo
Compressor summary: The paper introduces new sharpness measures for optimization algorithms in overparameterized models like neural networks, and shows their effectiveness in minimizing these measures and generalizing better.
https://openreview.net/forum?id=9U29U3cDKq
Compressor summary: The paper introduces APMD, an algorithm that adjusts payoff perturbations in games to achieve faster convergence to Nash equilibria.
https://openreview.net/forum?id=9Tq4L3Go9f
Compressor summary: The paper proposes a new algorithm, BAC, that uses a Blended Exploitation and Exploration operator to address underestimated Q-values in deep reinforcement learning and improve performance in continuous control tasks and real-world robot problems.
https://openreview.net/forum?id=9Rroj9GIOQ
Compressor summary: This paper introduces SPP, a method to efficiently fine-tune large language models by preserving their sparsity and structure while optimizing their weights.
https://openreview.net/forum?id=9QRcp2ubDt
Compressor summary: The paper proposes a fair and efficient algorithm for selecting candidates from biased groups to institutions while maximizing true utility.
https://openreview.net/forum?id=9PQnc6EWdL
Compressor summary: The paper combines mirror descent-based variational inference with Gaussian processes for fast and accurate few-shot classification, improving uncertainty estimation and convergence speed.
https://openreview.net/forum?id=9L7BZiTtJR
Compressor summary: The text discusses how overparameterization can improve neural network performance but may reduce the benefits of curriculum learning, and presents a theoretical analysis of this interaction in a 2-layer network example.
https://openreview.net/forum?id=9HdQr68Zyl
Compressor summary: The paper proposes a new method, Contrastive Distribution Methods (CDM), to evaluate open-domain text generation by leveraging the connection between increasing model parameters and improved performance.
https://openreview.net/forum?id=9HPoJ6ulgV
Compressor summary: The authors present the first polynomial transformer, enabling secure inference with homomorphic encryption on full transformers for various tasks, with results comparable to traditional models.
https://openreview.net/forum?id=9GbAea74O6
Compressor summary: The paper presents REST, a fast and memory-efficient graph-based method for real-time EEG signal analysis, achieving high accuracy in epileptic seizure detection and classification tasks.
https://openreview.net/forum?id=9GLvXGkUE2
Compressor summary: This paper studies how one-layer transformers with softmax attention learn linear function classes in structured data models with balanced or imbalanced feature vectors using gradient descent and reveals their learning dynamics.
https://openreview.net/forum?id=9DMMvMTDur
Compressor summary: The paper proposes reward model ensembles and EvIL method to improve imitation learning by addressing weak reward recovery and poor reward shaping issues.
https://openreview.net/forum?id=9CCoVyFuEp
Compressor summary: The authors propose a loss shaping constraints method for long-term time series forecasting that aims to balance average performance and error bounds at each step, improving the distribution of errors across the predicted window.
https://openreview.net/forum?id=9BrydUVcoe
Compressor summary: QuIP# is a weight-only post-training quantization method that achieves state-of-the-art compression results using the randomized Hadamard transform, vector quantization with $E_8$ lattice codebooks, and fine-tuning.
https://openreview.net/forum?id=9BWRs6XF8P
Compressor summary: The paper proposes NegUCB, a novel learning-based method for effective negotiation that uses contextual combinatorial multi-armed bandits to tackle exploration-exploitation dilemma and large action spaces in negotiation problems.
https://openreview.net/forum?id=9BGi9PEhNn
Compressor summary: The authors compare ConvNet and Vision Transformer models on various aspects beyond ImageNet accuracy, finding differences in types of mistakes, output calibration, transferability, and feature invariance, suggesting a need for more nuanced model selection.
https://openreview.net/forum?id=9ANyvRtFGa
Compressor summary: HexGen is a distributed inference engine for large language models that supports efficient deployment across diverse GPUs and reduces inference costs by adaptive scheduling.
https://openreview.net/forum?id=99jx5U81jx
Compressor summary: The study evaluates how well large language models can explain their reasoning and suggests two metrics based on counterfactual simulatability to measure explanation quality.
https://openreview.net/forum?id=99UFZV2VpU
Compressor summary: Our paper investigates the theoretical benefits of goal relabeling in reinforcement learning, leading to a new algorithm (GOALIVE) and a complexity measure (GOAL-BE).
https://openreview.net/forum?id=93gjGDwqim
Compressor summary: The paper proposes ReconBoost, a novel multi-modal learning method that alternates between updating fixed modalities to balance exploration and exploitation, addressing modality competition issues.
https://openreview.net/forum?id=91QmrfztSP
Compressor summary: The paper proposes a new knowledge distillation method using an auxiliary variable to better transfer knowledge from a teacher model to a student model, improving performance over existing methods.
https://openreview.net/forum?id=8ySQaphUYH
Compressor summary: WISER is a new method that uses weak supervision and supervised representation learning to predict personalized cancer drug responses using genomic data from patients.
https://openreview.net/forum?id=8xKGZsnV2a
Compressor summary: The text introduces AquaLoRA, a method for protecting image generation models from unauthorized use by embedding watermarks in their architecture.
https://openreview.net/forum?id=8viuf9PdzU
Compressor summary: SNPSE is a score-based method for Bayesian inference in simulator-based models that uses conditional score-based diffusion models and a sequential training procedure to reduce simulation cost.
https://openreview.net/forum?id=8uzBOVmh8H
Compressor summary: The paper proposes a new method to improve Jacobi decoding for LLM inference by refining the target model to predict the fixed point faster, achieving significant speedup with maintained quality.
https://openreview.net/forum?id=8tzjEMF0Vq
Compressor summary: The paper proposes a method to learn a mixture of reward models for reinforcement learning from human feedback, accounting for diverse human preferences and showing its effectiveness on small-scale and large-scale language models.
https://openreview.net/forum?id=8t8zBaGFar
Compressor summary: The paper introduces window search, a semantic search problem with numeric labels, and presents a modular tree-based framework that significantly speeds up its solution.
https://openreview.net/forum?id=8q4EPdjTLE
Compressor summary: The text discusses the benefits and risks of open-source generative AI, and proposes an AI openness taxonomy system to classify and assess its potential impacts on society.
https://openreview.net/forum?id=8onaVSFTEj
Compressor summary: HGCN2SP is a new model for solving 2SP problems that uses a hierarchical graph and reinforcement learning to select representative scenarios efficiently and accurately.
https://openreview.net/forum?id=8nd1yBRCDl
Compressor summary: EquiAV is a novel framework that uses equivariance for audio-visual contrastive learning, enabling robust supervision with minimal computational overhead and outperforming previous works.
https://openreview.net/forum?id=8mKXMnhnFW
Compressor summary: The paper proposes a novel zero-shot quantization approach that generates synthetic data based on the sharpness of the quantized model, improving generalization.
https://openreview.net/forum?id=8m4V6Fx6ma
Compressor summary: Key points: - The paper studies semi-supervised node classification on a synthetic dataset with labeled and unlabeled nodes - The paper finds the information threshold for exact recovery of test nodes using transductive learning - The paper proposes an optimal spectral estimator based on PCA and evaluates graph ridge regression and GCN Summary: The paper investigates semi-supervised node classification on a synthetic dataset with feature vectors derived from a Gaussian Mixture Model. It identifies the information threshold for exact recovery using transductive learning and proposes an optimal spectral estimator based on PCA. It also evaluates graph ridge regression and GCN and shows their potential to achieve the threshold.
https://openreview.net/forum?id=8l1KYguM4w
Compressor summary: Make-A-Shape is a fast and versatile 3D generative model that uses wavelet-tree representation to encode shapes and create various applications with minimal loss.
https://openreview.net/forum?id=8kLzL5QBh2
Compressor summary: Key points: - Transformers perform poorly in multivariate long-term forecasting compared to linear models - Attention mechanism is the main cause of this issue - SAMformer is a new lightweight transformer model that overcomes this problem with sharpness-aware optimization - SAMformer outperforms existing methods and is smaller than MOIRAI Summary: The paper introduces SAMformer, a transformer model that improves multivariate long-term forecasting by avoiding bad local minima with sharpness-aware optimization, and shows its superiority over linear models and other foundation models.
https://openreview.net/forum?id=8iWDWQKxJ1
Compressor summary: A coreset is a small subset of a set of points that can approximate the cost of median queries within a factor of $(1\pm\varepsilon)$, and its size depends on $\varepsilon$ and the dimension $d$.
https://openreview.net/forum?id=8iUgr2nuwo
Compressor summary: Wukong is a new recommendation model that achieves better performance and follows a scaling law, making it more efficient and adaptable to complex real-world datasets.
https://openreview.net/forum?id=8ho1l6RZNB
Compressor summary: The paper introduces image hijacks, which can manipulate the output and behaviour of vision-language models by controlling their image input with adversarial images.
https://openreview.net/forum?id=8h0x12p3zq
Compressor summary: This paper develops a method to measure how well graph neural networks generalize to new data when there are more parameters than data points, using an upper bound with a convergence rate of $O(1/n)$.
https://openreview.net/forum?id=8f8SI9X9ox
Compressor summary: The paper proposes new algorithms for dividing planar graphs into fair clusters with considerations of connectivity and optimality, and applies them to real networks modeling political redistricting.
https://openreview.net/forum?id=8dX4YnosqG
Compressor summary: The paper analyzes how numerical errors accumulate in PCA's deflation method when finding principal components sequentially using different algorithms or power iteration.
https://openreview.net/forum?id=8ZDFn7BDaH
Compressor summary: The paper proposes a novel hybrid pipeline combining asynchronous sensing and synchronous processing to enable classic processing of continuous ultra-sparse spatiotemporal data with dense machine learning models, achieving state-of-the-art performance and low latency.
https://openreview.net/forum?id=8Z2xWhuT6R
Compressor summary: The paper studies the theoretical separation between multimodal and unimodal learning, showing that average-case computational advantages of multimodal learning may be rare in practice due to their connection to cryptographic key agreement protocols.
https://openreview.net/forum?id=8WSNl2XA9r
Compressor summary: The authors propose a new metric (Delta Score) to evaluate molecular binding specificity in Structure-based Drug Design and develop an energy-guided generative model using contrastive learning that improves both specificity and traditional docking scores.
https://openreview.net/forum?id=8VEGkphQaK
Compressor summary: The paper proposes a synthetic graph navigation task to study how autoregressive Transformer models perform stepwise inference and explores various phenomena observed in these models.
https://openreview.net/forum?id=8STOjGCkfH
Compressor summary: HyperFields is a method for generating text-conditioned Neural Radiance Fields with a single forward pass, using a dynamic hypernetwork and NeRF distillation training to learn a general map between text and scenes.
https://openreview.net/forum?id=8RwhTPACAO
Compressor summary: The paper develops a general framework to study the distribution of the minimum empirical risk (MER) in statistical and machine learning problems, and shows how it can be used for inference and hypothesis testing with applications to neural networks.
https://openreview.net/forum?id=8PTx4CpNoT
Compressor summary: The paper shows that a language model of code can learn to represent the semantics of programs without explicit guidance, and introduces a new probing technique to study this phenomenon.
https://openreview.net/forum?id=8NfHmzo0Op
Compressor summary: Context-guided diffusion (CGD) is a method that uses unlabeled data and constraints to improve out-of-distribution generation for guided diffusion models in various application domains.
https://openreview.net/forum?id=8KeD4mEh3j
Compressor summary: HI-Mol is a novel method for data-efficient molecular generation that uses multi-level token embeddings and textual inversion to capture hierarchical information in molecule distribution.
https://openreview.net/forum?id=8JFIKpzumn
Compressor summary: The paper studies how multiple senders can persuade a self-interested agent using signaling, proposes a new game-theoretic solution concept, shows its computational hardness, and develops a neural network approach to find local equilibria in this setting.
https://openreview.net/forum?id=8GYclcxQXB
Compressor summary: The paper studies how time-continuous Markov jump processes on discrete state spaces relate to state-continuous diffusion processes, and proposes a training algorithm for the time-reversal of such processes using conditional expectations.
https://openreview.net/forum?id=8ERo4jph0A
Compressor summary: The paper proposes IGSG, a metric and regularization method to improve the robustness of classification over categorical inputs against adversarial attacks.
https://openreview.net/forum?id=8AeuhCgRRv
Compressor summary: The paper extends infinite-width analysis to Jacobians of deep neural networks, showing that MLPs and their Jacobians at initialisation converge to Gaussian processes in the infinite-width limit.
https://openreview.net/forum?id=89kZWloYQx
Compressor summary: The paper analyzes factors contributing to catastrophic forgetting in continual learning, showing that task sequence and algorithmic parameters affect forgetting, and validates the theory with simulations on linear regression models and DNNs.
https://openreview.net/forum?id=87ZrVHDqmR
Compressor summary: The paper introduces Med-ST, a framework for fine-grained spatial and temporal modeling of chest radiographs and radiological reports to exploit information from multiple views and temporal sequences.
https://openreview.net/forum?id=87CYNyCGOo
Compressor summary: Leddam is a learnable decomposition method and dual attention module for better multivariate time series forecasting, outperforming state-of-the-art methods.
https://openreview.net/forum?id=8296yUBoXr
Compressor summary: DIDI is a new method that learns diverse skills from offline data using diffusion probabilistic models as guides, achieving good results in decision-making tasks.
https://openreview.net/forum?id=7zvl9mNQG2
Compressor summary: The paper introduces Decomp Diffusion, an unsupervised method that decomposes images into compositional components, which can be flexibly combined to generate new scenes.
https://openreview.net/forum?id=7zEoinErzQ
Compressor summary: The paper investigates how deep neural networks learn and forget features through layers during forward propagation, and tracks the emergence and disappearance of interactions in each layer.
https://openreview.net/forum?id=7yixJXmzb8
Compressor summary: Finetuning pretrained models can introduce privacy risks due to backdoors that allow attackers to reconstruct data and bypass differential privacy guarantees.
https://openreview.net/forum?id=7xzhKEPfBo
Compressor summary: The text studies how to estimate risk measures like variance and VaR for an infinite-horizon Markov cost process, showing that it requires at least $\Omega(1/\epsilon^2)$ samples and deriving upper bounds for CVaR and variance.
https://openreview.net/forum?id=7wgXuNOF0V
Compressor summary: Boximator is a new approach for fine-grained motion control in video synthesis, using hard and soft boxes to define object positions and shapes, and self-tracking technique to learn box-object correlations.
https://openreview.net/forum?id=7uwLvFvpis
Compressor summary: The paper introduces a new self-supervised learning model, VQ-MTM, for EEG data analysis that uses vector quantization and phase alignment to learn robust features and outperforms existing methods on seizure detection and classification tasks.
https://openreview.net/forum?id=7tyAO5tUF8
Compressor summary: The paper proposes a new method to improve multi-plane representation in neural radiance fields by combining it with a coordinate-based MLP network that captures low-frequency details, resulting in better stability, efficiency, and performance.
https://openreview.net/forum?id=7sgqXa4aNM
Compressor summary: The paper introduces a general framework for learning from weak supervision (GLWS) with an EM-based algorithm that simplifies the computational demands and shows better performance and versatility across various scenarios.
https://openreview.net/forum?id=7rrN6E4KU0
Compressor summary: The study finds that pre-training on a public dataset improves differentially private learning by enhancing feature representation and suggests strategies like feature normalization and dimension reduction to improve robustness.
https://openreview.net/forum?id=7rfZ6bMZq4
Compressor summary: The authors propose a method called DoGE that optimizes how to sample training data from different domains to improve the generalization of large language models.
https://openreview.net/forum?id=7rTbqkKvA6
Compressor summary: The paper presents a new dataset and Symmetrized GNN model for predicting energy barriers in metallic glasses, achieving high accuracy and fast inference time compared to previous methods.
https://openreview.net/forum?id=7mFSaP6IiN
Compressor summary: This paper explores how integrating linear attention with speculative decoding can improve the efficiency and performance of autoregressive large language models, achieving significant reductions in perplexity and generation time.
https://openreview.net/forum?id=7joG3i2pUR
Compressor summary: OBAC is a new off-policy RL framework that adapts to a better offline policy to improve online learning, achieving high sample efficiency and performance on 53 tasks.
https://openreview.net/forum?id=7iH9RgMrzX
Compressor summary: The paper proposes a machine learning approach for predicting optimal dual solutions and an adaptive stabilization technique to improve the convergence rate of column generation.
https://openreview.net/forum?id=7gEcbhMqKU
Compressor summary: The paper introduces and analyzes stochastic proximal samplers for non-log-concave distributions, improving upon existing Langevin-based methods in terms of convergence and sample complexity.
https://openreview.net/forum?id=7emOSb5UfX
Compressor summary: The paper proposes an adaptive watermarking strategy for AI-generated text that balances quality, robustness, and security by adjusting the token distributions and using semantic embedding for output logits scaling.
https://openreview.net/forum?id=7dP6Yq9Uwv
Compressor summary: The paper proposes Dynalang, an agent that learns to understand diverse language, predict future states of the world, and act accordingly, outperforming existing methods on various tasks.
https://openreview.net/forum?id=7ckuC9C2FZ
Compressor summary: This paper introduces a new method for reconstructing surfaces from unorganized points using neural networks and global shape priors, which improves the reliability and quality of surface reconstruction.
https://openreview.net/forum?id=7bjyambg4x
Compressor summary: Maestro is a framework that trains low-rank DNNs with a novel low-rank ordered decomposition, allowing efficient compression and performance preservation.
https://openreview.net/forum?id=7bg10Jj3bG
Compressor summary: SDT is a new framework that uses signal temporal logic and Decision Transformer to train safer and more effective offline reinforcement learning policies for complex tasks.
https://openreview.net/forum?id=7XZKzQtooN
Compressor summary: The paper proposes two efficient algorithms for online low rank matrix completion recommendation problems and analyzes their performance under certain assumptions.
https://openreview.net/forum?id=7RSIGQRT1F
Compressor summary: The authors propose a Riemannian framework based on the Shahshahani metric to decompose games into simpler components called incompressible games, which have constant of motion and Poincaré recurrent properties in EW dynamics.
https://openreview.net/forum?id=7RHFdkAkVY
Compressor summary: The AttNS method improves the generalization and robustness of AI-Hybrid numerical solvers for differential equations by incorporating attention mechanisms inspired by ResNet.
https://openreview.net/forum?id=7R3pzxTSlg
Compressor summary: StructChem is a prompting strategy that improves GPT-4's chemical reasoning by providing an effective reasoning structure.
https://openreview.net/forum?id=7Qf1uHTahP
Compressor summary: The paper proposes a framework for learning optimal policies that balance long-term and short-term rewards, even when some long-term outcomes are missing, and shows its effectiveness through experiments.
https://openreview.net/forum?id=7PXSc5fURu
Compressor summary: FQI-LOG is a batch RL method that learns near-optimal policies with fewer samples by using log-loss and scaling costs with the optimal achievable cost.
https://openreview.net/forum?id=7OPHCeXcSS
Compressor summary: The paper proposes a new hypergradient method for constrained bilevel optimization without strict assumptions and a single-loop algorithm with a proven convergence rate.
https://openreview.net/forum?id=7KtFQnF368
Compressor summary: Inexact Riemannian gradient descent (RGD) can efficiently solve constrained nonconvex optimization problems within a general framework of tangential Block Majorization-Minimization (tBMM), with theoretical and experimental advantages over existing methods.
https://openreview.net/forum?id=7JKVPNEBkU
Compressor summary: The paper examines responsible AI licenses, which are used to manage risks of AI misuse, and suggests standardizing them while allowing some context-specific customizations.
https://openreview.net/forum?id=7E4c2gyP0R
Compressor summary: The paper proposes DiffFPR, a method that combines an iterative engine and a diffusion model to solve the challenging Fourier phase retrieval problem for multi-channel color images.
https://openreview.net/forum?id=7DbIyQlfaO
Compressor summary: The paper proposes using local intrinsic dimension (LID) of activations to measure and improve truthfulness in large language models for question answering tasks.
https://openreview.net/forum?id=7C4EQqtb02
Compressor summary: The paper presents a discrete graphon game model that simplifies stochastic games with heterogeneous interactions by using a representative player, and shows how to find and use its equilibrium solution.
https://openreview.net/forum?id=7AF0AMI4AE
Compressor summary: The text discusses the role of symmetry in neural network loss functions and its effects on model learning behavior, leading to different constraints and properties.
https://openreview.net/forum?id=76zq8Wkl6Z
Compressor summary: Next-token prediction models may fail due to autoregressive inference and teacher-forcing, leading to errors in training and in-distribution failure.
https://openreview.net/forum?id=761UxjOTHB
Compressor summary: Spectral DeTuning is a method that can recover the pre-fine-tuning weights of generative models, posing a new vulnerability to large-scale models like personalized Stable Diffusion and aligned Mistral.
https://openreview.net/forum?id=75Hes6Zse4
Compressor summary: The paper explores different ways to combine Evolutionary Algorithms and Reinforcement Learning for policy optimization, evaluates their effectiveness, and proposes new methods that achieve state-of-the-art results on various tasks.
https://openreview.net/forum?id=72oT4mPLUb
Compressor summary: The paper studies how self-attention models work like Markov chains, and how they learn from single output trajectories, explaining why large language models tend to produce repetitive text.
https://openreview.net/forum?id=71ktaA3ihI
Compressor summary: The paper presents a new sample compression technique for agnostic regression with different losses and explores its limits and open questions.
https://openreview.net/forum?id=70jplnkLMe
Compressor summary: PPFlow is a novel AI method for designing targeted peptide drugs that considers torsion angles and uses a new protein-peptide binding dataset for training deep learning models.
https://openreview.net/forum?id=6yQ5mIYxjj
Compressor summary: The paper proposes generalization bounds for algorithmic stability with unbounded loss functions and develops new concentration inequalities for subweibull random variables.
https://openreview.net/forum?id=6wVlH96oMX
Compressor summary: The paper proposes a method to improve Bayesian inference efficiency by using universal symmetries in joint models and penalizing violations with a self-consistency loss.
https://openreview.net/forum?id=6w7zkf9FBR
Compressor summary: The paper proposes a novel neural operator construction method using orthogonal attention, which improves overfitting and data efficiency for modeling PDE solutions.
https://openreview.net/forum?id=6pHP51F55x
Compressor summary: GeoAB is a novel approach for antibody design that addresses issues with structure authenticity and affinity maturation by using a generative geometry initializer and a position refiner, achieving state-of-the-art performance in co-design and mutation effect predictions.
https://openreview.net/forum?id=6n99bIxb3r
Compressor summary: This paper proposes methods for unsupervised learning in combinatorial optimization by constructing probabilistic objectives and derandomizing complex conditions using theoretical justification, achieving better optimization quality and speed.
https://openreview.net/forum?id=6kMMgmeM2U
Compressor summary: SelfVC uses self-synthesized examples to improve a voice conversion model with entangled speech representations derived from self-supervised learning and speaker verification models, achieving state-of-the-art results in zero-shot voice conversion.
https://openreview.net/forum?id=6jmdOTRMIO
Compressor summary: The paper proposes new debate protocols for AI safety that let humans simulate and verify stochastic AI systems with a polynomial number of steps.
https://openreview.net/forum?id=6h6ovHcC9G
Compressor summary: The paper presents a fast and memory-efficient algorithm for finding dense subgraphs in data mining and clustering, with similar quality to existing methods but faster speed and applicability to MPC.
https://openreview.net/forum?id=6djDWVTUEq
Compressor summary: Subgraphormer combines Subgraph GNNs and Graph Transformers to improve expressive power, message-passing mechanisms, and aggregation schemes for graph neural networks, using attention and positional encodings based on the product graph.
https://openreview.net/forum?id=6dKUu2EkZy
Compressor summary: The paper proposes efficient algorithms for inverse reinforcement learning using polynomial samples and runtime, with theoretical guarantees and near-optimal sample complexities.
https://openreview.net/forum?id=6axTFAlzRV
Compressor summary: FedLESAM is a novel algorithm that improves and speeds up federated learning by estimating global perturbations locally and mitigating sharpness issues in heterogeneous environments.
https://openreview.net/forum?id=6aKwVmHQI1
Compressor summary: This paper proposes a method to train privacy-preserving vision models using self-supervised learning with differential privacy, achieving comparable performance to non-private models on standard tasks.
https://openreview.net/forum?id=6Zl9rv6PDx
Compressor summary: CAIAC is a data augmentation method that uses counterfactual reasoning to create synthetic transitions from a fixed dataset, improving offline robot learning by increasing its robustness against distributional shift.
https://openreview.net/forum?id=6Zgjrowepn
Compressor summary: The paper proposes a new fairness notion for federated learning called *PVC-stability*, which is based on rank rather than utility, and presents an algorithm called Rank-Core-Fed that achieves this fairness goal.
https://openreview.net/forum?id=6XH8R7YrSk
Compressor summary: The paper investigates the differences between reward-based (PPO) and reward-free (DPO) methods for aligning large language models with human preferences, and shows that PPO performs better in various tasks.
https://openreview.net/forum?id=6Wauue8pWd
Compressor summary: The paper introduces multicalibration, a method for generating reliable confidence scores for outputs of large language models by calibrating across intersecting data groupings.
https://openreview.net/forum?id=6WYk5R86Wl
Compressor summary: The authors present an efficient method to train and predict using Gaussian factor graphs, which can handle deep networks and continual learning tasks.
https://openreview.net/forum?id=6VZOONPn8S
Compressor summary: The paper shows that linearizing non-linear activations in neural networks can reduce accuracy for minority groups and suggests a mitigation strategy.
https://openreview.net/forum?id=6VQXLUy4sQ
Compressor summary: Langevin Predictive Coding (LPC) is a new algorithm that improves deep generative model learning using Gaussian noise injection and encoder network initialization, resulting in better sample quality, faster convergence, and comparable or higher performance than VAEs on key metrics.
https://openreview.net/forum?id=6UGSDDPkJw
Compressor summary: The paper reviews popular methods of explaining computer vision models and proposes new research directions for using contextual information to improve explanations.
https://openreview.net/forum?id=6TM62kpI5c
Compressor summary: The paper proposes FedGF, a method to improve generalization in federated learning by reducing flatness discrepancy and pursuing flatter minima of global models.
https://openreview.net/forum?id=6TCeizkLJV
Compressor summary: This paper proposes an ICRL method that can estimate constraints from expert demonstrations with a specified confidence level and helps users decide if more data is needed to achieve both reliable constraints and good performance.
https://openreview.net/forum?id=6PqWuSuWvX
Compressor summary: The text introduces a method to compromise language models via covert malicious finetuning that evades detection by common defense mechanisms.
https://openreview.net/forum?id=6PTiCmGcNx
Compressor summary: Key points: - Anomaly detection methods often need large normal datasets, but this is unrealistic for many applications. - The paper proposes to use two small additional datasets with partially-observed normal and anomaly samples to learn the normal distribution from contaminated data. - The paper proves that the method works under some conditions, considers overfitting issue, and shows experimental results. Summary: The paper presents a new anomaly detection method that learns from two small datasets with partial observations of normal and anomalous samples, and proves its correctness and effectiveness.
https://openreview.net/forum?id=6P88DMUDvH
Compressor summary: Code as Reward (VLM-CaR) is a framework that generates reward functions from Vision-Language Models for faster training of Reinforcement Learning agents.
https://openreview.net/forum?id=6OkvBGqW62
Compressor summary: HypDiff is a novel framework that uses hyperbolic geometry to create an anisotropic latent space for graph generation, preserving the original topological properties.
https://openreview.net/forum?id=6OSLjErBhh
Compressor summary: The paper shows how probabilistic inference can help estimate total variation distance between same-structure distributions in Bayes nets using partial couplings.
https://openreview.net/forum?id=6NQ77Vj3DT
Compressor summary: The authors develop a fast gradient-based learning method for neural-symbolic systems using convex and bilevel optimization, which improves the inference speed and prediction performance.
https://openreview.net/forum?id=6L4K5jmSJq
Compressor summary: The paper explores whether there are completely parameter-free methods for stochastic optimization, showing that simple hyperparameter search can achieve this in some cases, while proving a lower bound for convex optimization.
https://openreview.net/forum?id=6KtXzUUEp4
Compressor summary: The paper presents a method to adjust machine learning models after training to ensure fairness across multiple groups and applies it to various scenarios like image segmentation, classification, and text generation.
https://openreview.net/forum?id=6Kg9p8URlj
Compressor summary: The study derives the first meaningful generalization bounds for pretrained large language models, showing they can discover regularities beyond their training data, and develops a simple parameterization method to achieve this.
https://openreview.net/forum?id=6KLNiRdWH6
Compressor summary: The Sinkhorn Method of Moments is an optimal transport-based instrumental variable estimator that improves robustness against data corruption and adversarial attacks by using data-derivative information to capture the geometry of the data manifold.
https://openreview.net/forum?id=6FtAXU4ean
Compressor summary: The paper introduces a new evaluation protocol for Test Time Adaptation (TTA) methods that considers their adaptation speed and shows that faster, simpler methods can outperform slower, more sophisticated ones.
https://openreview.net/forum?id=6FXtu8clyp
Compressor summary: The text discusses the growing use of visually-conditioned language models (VLMs) in various applications, their key design decisions, and a new framework for evaluating them, along with improved VLM models that outperform existing ones.
https://openreview.net/forum?id=6EF0bxcZvT
Compressor summary: The paper proposes online learning algorithms that balance predictive accuracy and individual fairness, improving on existing bounds and reducing oracle calls.
https://openreview.net/forum?id=6DBvBcW770
Compressor summary: The paper proposes a method called Stochastic Resonance Transformer (SRT) that improves ViT models by applying sub-token spatial transformations and aggregating the features, resulting in finer-scale spatial information without fine-tuning.
https://openreview.net/forum?id=6D0nyemiWk
Compressor summary: The paper explores different graphical and analytical methods to identify causal effects in non-positive and unconfounded observational studies, and proposes a new algorithm based on these approaches.
https://openreview.net/forum?id=6CV1N7hhpA
Compressor summary: The paper introduces non-positive kernels for GP regression that avoid mean reversion and its issues, while maintaining desirable smoothness and stationarity properties.
https://openreview.net/forum?id=6BYD121JFO
Compressor summary: The paper proposes a new method to compress grid-based NeRF models, using neural compression and importance-weighted rate-distortion, achieving better efficiency and quality than previous methods.
https://openreview.net/forum?id=69RewQwWA9
Compressor summary: The paper proposes a new sparse Bayesian learning algorithm that guarantees global convergence and improves performance in various applications.
https://openreview.net/forum?id=685vj0lC9z
Compressor summary: The study examines how large language models balance honesty and helpfulness in communication based on human preferences and feedback methods.
https://openreview.net/forum?id=66k81s33p3
Compressor summary: The paper proposes a new method, EXO, to optimize language models' policies based on human preferences more efficiently than existing methods like DPO and RL.
https://openreview.net/forum?id=66KmnMhGU5
Compressor summary: Inner Interpretability is an emerging field that seeks to understand the inner workings of AI systems, facing similar challenges as Cognitive Neuroscience, which it can learn from to develop better mechanistic explanations.
https://openreview.net/forum?id=65XKBGH5PO
Compressor summary: The authors present a method for reconstructing visual perception from brain activity using only 1 hour of fMRI training data and achieve high-quality results, outperforming previous single-subject approaches.
https://openreview.net/forum?id=64fdhmogiD
Compressor summary: The paper explores how to achieve stability in reinforcement learning, especially for unbounded state spaces, by using a Lyapunov-based cost-shaping technique and state transformations.
https://openreview.net/forum?id=64MQCia06B
Compressor summary: The paper presents a new deep learning method for simulating Schrödinger equation that adapts to low-dimensional structure of wave function, reducing computational complexity and outperforming existing methods.
https://openreview.net/forum?id=64I29YeQdt
Compressor summary: RefQD is a novel method that improves resource efficiency in Quality-Diversity optimization by decomposing neural networks into representation and decision parts, sharing the representation among decision parts, and addressing mismatch issues.
https://openreview.net/forum?id=61WtHsVKWF
Compressor summary: The paper explores the trade-offs between competitive ratios and consistency in online bipartite matching algorithms with different arrival models and advice quality.
https://openreview.net/forum?id=61RlaY9EIn
Compressor summary: Key points: - Machine learning technologies rely on large datasets, but pose privacy risks for people's data. - Existing solutions have limitations in terms of data quality, availability, or theory. - The paper proposes a formal definition and a learnable data transformation framework to protect privacy while preserving utility. - The method is evaluated on various datasets and tasks, showing effectiveness and generalizability. Summary: The paper presents a novel data transformation framework that uses information theory to protect privacy and preserve utility of machine learning datasets, and shows its performance on different types of data and tasks.
https://openreview.net/forum?id=61JD8wp4Id
Compressor summary: The alignment of neural tangent kernels and data eigenvectors can improve convergence and generalization for graph neural networks, especially when using cross-covariance instead of just input covariance.
https://openreview.net/forum?id=61A1bsVjRg
Compressor summary: The paper proposes a novel algorithm called Tilt and Average that adjusts the final layer weights of neural networks to improve calibration and reliability of predictions.
https://openreview.net/forum?id=60vx5AfM3C
Compressor summary: The study investigates the geometric properties of sampled gradients in neural network optimization, finding predictable and consistent behavior that allows for theoretical guarantees of linear convergence and practical learning rate schedules.
https://openreview.net/forum?id=60vC1FY0dZ
Compressor summary: This paper shows that gradient regularization can cause problems in adaptive optimization scenarios and proposes three warmup strategies to improve performance, especially for scalable models.
https://openreview.net/forum?id=60HydCpCMZ
Compressor summary: The Riemannian Diffusion Mixture is a new generative diffusion model on manifolds that uses a mixture of bridge processes and doesn't require heat kernel estimations, enabling better performance and scalability on diverse geometries.
https://openreview.net/forum?id=60F0fVbknK
Compressor summary: The paper introduces DHT-CIT, a novel algorithm that learns causal relations from subsampled time series using only two time-slices, instead of full interventions.
https://openreview.net/forum?id=5zXTwX92qv
Compressor summary: BECoTTA is a framework for continuous test-time adaptation that uses Mixture-of-Domain Low-rank Experts to efficiently update model parameters for changing environments while maintaining generalization.
https://openreview.net/forum?id=5x788rqbcj
Compressor summary: This paper shows that large language models need diverse pretraining data for reliable knowledge extraction and suggests rewriting data and adding finetuning data as recommendations.
https://openreview.net/forum?id=5vZzmCeTYu
Compressor summary: The study compares various off-policy RL techniques on diverse simulation tasks and identifies consistent combinations that lead to robust performance improvements.
https://openreview.net/forum?id=5uwBzcn885
Compressor summary: Patchscopes is a framework that uses a large language model to explain its own internal representations in natural language, improving on existing interpretation methods and enabling new possibilities.
https://openreview.net/forum?id=5tPB5VXo87
Compressor summary: PepFlow is a new generative model for designing full-atom peptides that target specific protein receptors by learning the structure using rigid backbone frames, side-chain angles, and categorical distributions of residue types.
https://openreview.net/forum?id=5t4V7Q6lmz
Compressor summary: The text introduces a new framework, $\partial$-CROWN, that provides guarantees on the worst-case residual error of Physics-Informed Neural Networks (PINN) over their continuous applicability domain.
https://openreview.net/forum?id=5sgkNtexs2
Compressor summary: The paper proposes a novel raw image compression method that selectively compresses the noise-free part, discards real noise, and outperforms existing techniques with significant improvements in rate-distortion balance and bit saving.
https://openreview.net/forum?id=5pg9YJBaiG
Compressor summary: CTAG is a text-to-audio method that uses a modular sound synthesizer with 78 parameters for easy tweaking and inspection of high-quality, abstract sounds based on natural language prompts.
https://openreview.net/forum?id=5nxIRQ8GNa
Compressor summary: SPARC is a method to pretrain multimodal representations from image-text pairs using sparse similarity and fine-grained sequence-wise loss, which improves performance on various tasks and models.
https://openreview.net/forum?id=5nuW5iBAJS
Compressor summary: The study uses AI to predict the shape and size of nanoparticles from text descriptions, creating a new dataset and evaluating different models for this task.
https://openreview.net/forum?id=5mCaITRTmO
Compressor summary: The paper proposes AQLM, a method that compresses large language models with very low bit counts using learned additive quantization and joint optimization of codebook parameters, achieving high accuracy-to-size trade-off and outperforming existing schemes in extreme compression.
https://openreview.net/forum?id=5lI9wm4dws
Compressor summary: Key points: - The text proposes a new causal effect estimator under networked interference using targeted learning and neural networks - The estimator achieves double robustness and faster convergence than single nuisance models - The text provides theoretical analysis and experimental results to support the proposal Summary: The text introduces a novel doubly robust causal effect estimator under networked interference, using targeted learning and neural networks, which outperforms single nuisance models in theory and practice.
https://openreview.net/forum?id=5kXNMDpUVF
Compressor summary: The text proposes an adaptive hyperparameter optimization method for differentially private deep learning that achieves state-of-the-art performance on various tasks and privacy levels.
https://openreview.net/forum?id=5kVgd2MwMY
Compressor summary: SPO is a simple algorithm for reinforcement learning from human feedback that uses self-play, preference optimization, and minimax winner concept to handle non-Markovian, intransitive, and stochastic preferences efficiently and robustly.
https://openreview.net/forum?id=5kGfm3Pa41
Compressor summary: Key points: - Paper proposes new architecture for graph neural networks based on deep state-space models - Model aggregates nodes by their distances to target and uses linear RNN to encode hop representations - No positional encoding needed, better performance than graph transformers, lower computational cost Summary: The paper introduces a novel graph neural network architecture that uses deep state-space models to aggregate nodes based on their distances and a diagonal linear RNN to encode the information, outperforming graph transformers with less computation.
https://openreview.net/forum?id=5j7Lq2ASiU
Compressor summary: The text introduces new distributed bilevel optimization algorithms that reduce communication overhead by using compression techniques, while maintaining efficiency and scalability.
https://openreview.net/forum?id=5hfvLBgnNE
Compressor summary: The paper proposes matrix information theory tools to analyze and optimize information interactions in supervised learning processes.
https://openreview.net/forum?id=5cm2jGct2W
Compressor summary: The paper proposes a new fairness metric, MCDP, to measure local disparity in machine learning algorithms and develops optimization algorithms to improve fairness-accuracy trade-offs.
https://openreview.net/forum?id=5ap1MmUqO6
Compressor summary: The paper proposes a method to compress cross-view representation for partial multi-view multi-label learning, enhancing task-relevant information and label correlation learning.
https://openreview.net/forum?id=5ZwEifshyo
Compressor summary: Self-repair in large language models occurs when ablating attention heads, but it is imperfect and noisy due to varying mechanisms such as LayerNorm changes and Anti-Erasure.
https://openreview.net/forum?id=5WnKLIAX4q
Compressor summary: The paper introduces a new method for robust Gaussian process regression that preserves closed-form conditioning and is efficient in computation.
https://openreview.net/forum?id=5WEIVj98Ju
Compressor summary: The text proposes a new method to deal with distribution shifts in unsupervised domain adaptation by using an importance weighted group accuracy estimator for model calibration and selection tasks.
https://openreview.net/forum?id=5ToHnqYxjB
Compressor summary: The paper proposes ISA, a method to enhance the interpretability of deep neural networks by iteratively generating high-quality samples and distinguishing important features during gradient ascent and descent.
https://openreview.net/forum?id=5SpjhZNXtt
Compressor summary: The paper calls for using large generative models to develop automated systems that can discover scientific knowledge from existing datasets without needing more data or experiments, but acknowledges current limitations and suggests integrating tools and user feedback for better results.
https://openreview.net/forum?id=5S8ukkEQr2
Compressor summary: This paper introduces a new way to analyze regret in risk-sensitive reinforcement learning with hindsight observations, and provides an efficient algorithm that performs well even when the environment is risky or uncertain.
https://openreview.net/forum?id=5QWKec0eDF
Compressor summary: DivBS is a reference-model-free method that efficiently selects diverse and representative samples for machine learning models by measuring group-wise orthogonalized representativeness, achieving better performance-speedup trade-offs than previous methods.
https://openreview.net/forum?id=5PqzKxmfag
Compressor summary: Key points: - Deep neural networks struggle with robustness to spatially transformed inputs - Current approaches have limitations in data variability or inductive biases - Inspired by human perception, a new technique called ITS is proposed - ITS emulates mental or physical actions during inference using energy-based evaluations - ITS improves robustness to spatially transformed inputs without explicit biases or data augmentation Summary: The authors propose ITS, a novel inference method for deep neural networks that mimics human perception and enhances robustness to spatially transformed inputs by traversing a sparsified transformation tree during inference.
https://openreview.net/forum?id=5Pcl5qOOfL
Compressor summary: The text introduces a new algorithm that improves the reliability and efficiency of captioning unlabeled 3D objects using vision language models, and shows its usefulness for conditional inference and visual reasoning ablation.
https://openreview.net/forum?id=5PQhu8flSO
Compressor summary: The text discusses the selective inclusion of data points in sequential data, which can distort statistical analysis but also offer insights into hidden generation processes, and proposes a method to identify selection structures and dependencies in such data.
https://openreview.net/forum?id=5NTTCCO74S
Compressor summary: This paper proposes a new framework called "regression with deferral," which involves asking multiple experts for predictions and introduces new loss functions and consistency guarantees for this problem.
https://openreview.net/forum?id=5M4Qa9AqY7
Compressor summary: The paper introduces a method called Differentiable Adjacency Test (DAT) to learn causal graphs from large scale data using neural networks, which improves prediction accuracy for interventions.
https://openreview.net/forum?id=5JrlywYHRi
Compressor summary: Decor is a decentralized learning method that uses randomness seeds and correlated noises to protect users' privacy while maintaining optimal trade-offs between privacy and utility.
https://openreview.net/forum?id=5ILo43JIzg
Compressor summary: The paper proposes a new codebook method for learning image representations, inspired by Kepler's Conjecture, which improves generation and reconstruction tasks on various datasets.
https://openreview.net/forum?id=5ExWEazod5
Compressor summary: PEG is a method that samples and initializes Vision Transformers with elastic scales based on a probabilistic mixture approach, preserving their knowledge and adapting to different resource constraints.
https://openreview.net/forum?id=59oXyDTLJv
Compressor summary: The paper presents relaxed group convolutions as a flexible technique to learn asymmetries in data and uncover interpretable symmetry-breaking factors in various physical systems.
https://openreview.net/forum?id=59MYoLghyk
Compressor summary: BEAG is a graph construction method for goal-conditioned RL that efficiently explores subgoals and avoids unattainable ones using adaptive grid refinement.
https://openreview.net/forum?id=55HfvJ6lDB
Compressor summary: The paper proposes a new method for forecasting hierarchical time series that learns the optimal oblique projection from data, which improves accuracy and adaptability compared to existing methods.
https://openreview.net/forum?id=54NSHO0lFe
Compressor summary: SparseTSF is a lightweight model that simplifies long-term time series forecasting by decoupling periodicity and trend, using fewer than 1k parameters and generalizing well with limited resources.
https://openreview.net/forum?id=53iSXb1m8w
Compressor summary: This paper investigates the cause of poor transfer in fine-tuning RL models and shows that standard knowledge retention techniques can mitigate it, leading to better performance in challenging environments.
https://openreview.net/forum?id=5353dJE9Ek
Compressor summary: PAGER is a framework that uses both uncertainty and non-conformity scores to detect and characterize failures in deep regression models.
https://openreview.net/forum?id=51iwkioZpn
Compressor summary: The paper introduces CPO, a novel method to improve machine translation for moderate-sized large language models by avoiding adequate but not perfect translations, achieving performance comparable to or better than state-of-the-art models.
https://openreview.net/forum?id=51gXk4BISH
Compressor summary: The authors propose a novel method to model customer demand in online dynamic pricing problems with feature-based price elasticity and show that their Pricing with Perturbation (PwP) algorithm achieves near-optimal regret, while offering insights for practical pricing strategies.
https://openreview.net/forum?id=50vc4HBuKU
Compressor summary: QIREN is a quantum version of Fourier Neural Networks that can better represent high-frequency components in signals, leading to improved performance in various tasks.
https://openreview.net/forum?id=4zOZ0yKhm6
Compressor summary: The authors propose a probabilistic model that captures coordination between agents as temporal influence and show that it predicts team performance in a virtual search and rescue mission using speech and semantics.
https://openreview.net/forum?id=4zN9tvZfns
Compressor summary: PrivPGD is a new method for creating private data synthesis of protected tabular datasets, using optimal transport and particle gradient descent, which works better than current methods and can handle extra constraints.
https://openreview.net/forum?id=4zAHgkiCQg
Compressor summary: Large language models are sensitive to the order of premises in reasoning tasks, which can cause a significant drop in performance if the order does not match the context required for intermediate steps.
https://openreview.net/forum?id=4ye2I5OelI
Compressor summary: The paper studies how many samples are needed to learn Nash Equilibrium policies in Mean-Field Games using model-based RL with strategic exploration, and proposes a new complexity measure and algorithm that make the problem easier than previously thought.
https://openreview.net/forum?id=4uTJfGYA2t
Compressor summary: The paper presents a novel ASR framework for dialog systems that adapts to conversational context and user feedback using student-teacher learning, context-aware processing, and contrastive self-supervision with Ohm.
https://openreview.net/forum?id=4sikyurTLX
Compressor summary: The paper proposes a new framework called sample-specific multi-channel masks (SMM) to improve visual reprogramming by generating customized masks for each input image instead of using a shared and pre-defined one.
https://openreview.net/forum?id=4qsduFJDEB
Compressor summary: The N-particle underdamped Langevin algorithm is a new method for optimizing non-linear functionals related to mean-field neural networks and other problems, with improved convergence guarantees.
https://openreview.net/forum?id=4pFgOzKF76
Compressor summary: The paper studies online learning algorithms with limited memory, shows that common approaches have high regret, and proposes a better algorithm that depends on the order of past rewards.
https://openreview.net/forum?id=4oD0tRrUOX
Compressor summary: $\Phi_ extrm{Flow}$ is a Python toolkit that simplifies writing differentiable simulation code across different ML libraries and provides many advanced features for scientific applications.
https://openreview.net/forum?id=4mU6LNMaIu
Compressor summary: The paper proposes GroupCover, a new obfuscation method for DNN models that uses randomization and mutual covering to protect against model-stealing attacks and improves security over existing solutions.
https://openreview.net/forum?id=4lghifYrSU
Compressor summary: The paper proposes an online hard thresholding algorithm for contextual bandits with knapsack constraints, which exploits sparsity to reduce regret in high-dimensional settings.
https://openreview.net/forum?id=4jqOV6NlUz
Compressor summary: The paper proposes a new method to measure and improve Retrieval-Augmented Large Language Models' accuracy by generating synthetic exams based on task-specific documents and using Item Response Theory.
https://openreview.net/forum?id=4iy0q0carb
Compressor summary: Weighted frames improve equivariant neural networks by preserving function continuity and avoiding discontinuity issues caused by unweighted frame-averaging.
https://openreview.net/forum?id=4iBJyJeBX5
Compressor summary: The paper explores relaxing assumptions in variational inequality problems, studies structured non-monotone generalized smoothness, and shows convergence results for three methods using adaptive stepsizes.
https://openreview.net/forum?id=4dxR7awO5n
Compressor summary: This paper explores the relationship between memorization, differential privacy, and input loss curvature in deep neural networks and provides both theoretical and empirical evidence.
https://openreview.net/forum?id=4dOJAfXhNV
Compressor summary: The paper introduces SAPG, a new on-policy RL algorithm that improves performance in large-scale parallelized environments by chunking and importance sampling.
https://openreview.net/forum?id=4byOXWrJay
Compressor summary: The paper proposes a new GPLVM model that improves kernel flexibility and projection noise to avoid model collapse and achieve better latent representations and missing data imputation.
https://openreview.net/forum?id=4boDu42RtE
Compressor summary: The paper proposes $E^3$-FaceNet, a fast and accurate network for text-to-3D face generation and manipulation that uses a direct mapping from text to 3D visual space and enhances semantic alignment and geometric consistency.
https://openreview.net/forum?id=4axAQHwBOE
Compressor summary: Rob-FCP is a novel framework that provides robust federated conformal prediction in Byzantine settings, effectively countering malicious clients and preserving coverage guarantees.
https://openreview.net/forum?id=4ZrppmS42b
Compressor summary: LEVI is a novel method to improve fine-tuning generalization by adaptively ensembling pre-trained and task-specific models layer-wise, addressing limitations in both data sources.
https://openreview.net/forum?id=4Zr7T6UrBS
Compressor summary: The paper proposes a new method for offline learning, called Primal Wasserstein DICE, which uses a contrastively learned distance metric to minimize the gap between the learner's and expert's state occupancies.
https://openreview.net/forum?id=4XxsheIbtn
Compressor summary: Uni-Med is a framework for interpreting medical images using instructions and generating visual explanations, improving Med-VQA accuracy.
https://openreview.net/forum?id=4XlGXIh2BB
Compressor summary: The text discusses two contrasting views on the ethical agency of AI and its implications for system design and accountability.
https://openreview.net/forum?id=4Vqr8SRfyX
Compressor summary: The text discusses how large language models struggle with simple math problems like addition, relying on case-based reasoning instead of rule-based reasoning. The authors propose a technique called Rule-Following Fine-Tuning (RFFT) to teach LLMs to use rules and improve their generalization ability.
https://openreview.net/forum?id=4UWjqrMmFp
Compressor summary: The paper presents new coreset constructions for multiple $\ell_p$ regression and related problems with optimal or near-optimal size and approximation guarantees.
https://openreview.net/forum?id=4RqG4K5UwL
Compressor summary: KEP-SVGP is a method to estimate uncertainty in self-attention models using asymmetric kernels and reduced complexity.
https://openreview.net/forum?id=4PuM6iGPPi
Compressor summary: The paper proposes a theoretical framework and an efficient network training approach called Deep Fusion to reduce the cost of training large language models while maintaining performance.
https://openreview.net/forum?id=4PB1RMsUy4
Compressor summary: The authors propose a new spiking mechanism for language tasks that generalizes better than existing methods and reduces the performance gap between spiking neural networks and artificial neural networks.
https://openreview.net/forum?id=4KQ0VwqPg8
Compressor summary: The paper proposes max-reward RL, an approach to learn from rewards that optimizes the maximum reward instead of the cumulative one, and shows its advantages in goal-reaching tasks over standard RL.
https://openreview.net/forum?id=4HCi7JGCZk
Compressor summary: The paper proposes a size-invariant evaluation method for salient object detection that improves the performance in detecting objects of different sizes.
https://openreview.net/forum?id=4G5Dcjcm1s
Compressor summary: The paper introduces VinT-6D, a large dataset for object-in-hand pose estimation using vision, touch, and proprioception, and presents a benchmark method that fuses multi-modal information to improve robotic manipulation.
https://openreview.net/forum?id=4FJJfYjUQR
Compressor summary: The paper proposes improved graph transformer architectures aligned with the Weisfeiler--Leman hierarchy, achieving better expressivity and practicality, as well as studying positional encodings and testing on a large molecule dataset.
https://openreview.net/forum?id=4DAl3IsvlU
Compressor summary: The authors establish a link between network geometry and causal inference, showing that negative curvature can hinder estimating causal parameters and proposing a method using geometric Ricci flow to reduce estimation error in networked data.
https://openreview.net/forum?id=4CO45y7Mlv
Compressor summary: The study shows that using conformal prediction sets, which provide alternative answers with varying levels of certainty, improves human accuracy in decision making tasks compared to fixed-size prediction sets.
https://openreview.net/forum?id=4BWCecFEcQ
Compressor summary: The text proposes PerceptAnon, a learning-based metric that evaluates the privacy of anonymized images based on human perception of anonymity and contextual backgrounds, and introduces a curated dataset for assessing image anonymization.
https://openreview.net/forum?id=4BIOZSz7zU
Compressor summary: This paper investigates non-Markovian fairness, where multiple stakeholders are affected by sequential decision making processes that depend on history and need to be assessed at different time points.
https://openreview.net/forum?id=49vHLSxjzy
Compressor summary: The paper proposes a probabilistic method to learn the degree of equivariance in steerable convolutional neural networks, which model geometric symmetries and can handle mixed symmetries.
https://openreview.net/forum?id=47jMS97wJX
Compressor summary: E-UC$^3$RL is an algorithm for regret minimization in stochastic CMDPs with efficient and rate-optimal performance under minimal assumptions and general offline function approximation setting.
https://openreview.net/forum?id=47ahBl70xb
Compressor summary: This paper explores the expressive power of deep neural networks combining linear RNNs and MLPs, showing that using complex numbers in the recurrence can improve information storage and help with long-range reasoning tasks.
https://openreview.net/forum?id=46vXhZn7lN
Compressor summary: The paper proposes using multilevel Monte Carlo to improve nested Bayesian optimization methods, achieving better convergence rates and avoiding smoothness assumptions.
https://openreview.net/forum?id=45HNimd4YI
Compressor summary: The text describes new efficient algorithms for private data release, especially for cosine similarities and high-dimensional marginal queries, with improved guarantees for sparse datasets and theoretical explanation for their effectiveness.
https://openreview.net/forum?id=44qxX6Ty6F
Compressor summary: This paper proposes using randomness in machine learning algorithms to fairly distribute scarce resources, considering individual claims and interests.
https://openreview.net/forum?id=43HZG9zwaj
Compressor summary: Diffusion tempering is a novel technique that improves gradient-based optimization of parameters in ordinary differential equations by reducing noise in probabilistic numerical methods.
https://openreview.net/forum?id=40hCy8n5XH
Compressor summary: The paper introduces InfoNet, a neural network that efficiently estimates mutual information between data streams using attention mechanism and deep learning infrastructures, enabling real-time applications like test-time optimization of neural networks or end-to-end learning.
https://openreview.net/forum?id=40foON48am
Compressor summary: The paper studies matrix completion with constraints and provides sample complexity bounds, a new weighted trace norm, and a non-linear model (FRMC) that improves performance on real data.
https://openreview.net/forum?id=3xPMW9JURD
Compressor summary: The paper presents a new framework for learning neural network (NN) controllers with Lyapunov stability guarantees using fast empirical falsification and strategic regularizations, without relying on expensive solvers for stability verification.
https://openreview.net/forum?id=3umNqxjFad
Compressor summary: The paper presents a modified A3C algorithm that uses ReLU, spectral normalization, dropout, and Thompson sampling to improve approximate Bayesian inference for deep reinforcement learning.
https://openreview.net/forum?id=3uPSQmjXzd
Compressor summary: Key points: - The paper proposes a representation learning method to deal with dynamics mismatch in RL - The method measures and penalizes representation deviations from the source domain as a reward signal - The method shows strong performance on various tasks with kinematic and morphology mismatch Summary: The paper presents a novel decoupled representation learning approach for RL that adapts to dynamics mismatch by using representation deviation as a reward penalty, and demonstrates its effectiveness on different tasks.
https://openreview.net/forum?id=3tJDnEszco
Compressor summary: The authors present an AI-guided framework that combines linguistic reasoning with quantum-chemistry based feedback to discover new and efficient catalysts for sustainable chemical processes.
https://openreview.net/forum?id=3pxMIjB9QK
Compressor summary: The paper introduces and studies the biharmonic distance, a measure of edge importance for graph topology, and develops algorithms based on it.
https://openreview.net/forum?id=3o7G6tIo4X
Compressor summary: The paper analyzes overfitting in deep weight space models and proposes a MixUp method for data augmentation to improve performance in classification and contrastive learning tasks.
https://openreview.net/forum?id=3nlBesNxcm
Compressor summary: The paper proposes efficient algorithms for multivariate linear regression with missing data, using $L_2$ and $L_1$ loss functions and penalties, and provides rigorous error bounds and experimental results.
https://openreview.net/forum?id=3mQ6ZKTSQl
Compressor summary: The paper shows that prompting and prefix-tuning can universally approximate sequence-to-sequence functions with smaller pretrained models, especially using attention mechanisms.
https://openreview.net/forum?id=3hSTecKy1b
Compressor summary: The text discusses challenges in training foundation models due to data collection issues and suggests using universal data provenance standards for more ethical and trustworthy development.
https://openreview.net/forum?id=3eHNvPHL9Z
Compressor summary: The paper investigates why over-parameterized neural networks, trained to perfectly fit the data, generalize well and shows that it is because a flat prior over the network parameters induces a rich prior over the functions, leading to simpler functions that require fewer parameters.
https://openreview.net/forum?id=3d5CIRG1n2
Compressor summary: The paper proposes DoRA, a method that combines LoRA and weight decomposition to improve fine-tuning of language models without increasing inference costs.
https://openreview.net/forum?id=3ash2ksk1r
Compressor summary: The paper analyzes Wasserstein GANs, which generate diverse examples without replication or deviating from the empirical distribution, while maintaining statistical optimality.
https://openreview.net/forum?id=3ajK5xplDL
Compressor summary: The paper presents a modified 20-year-old algorithm that can perform private clustering under various privacy models, including a new one for continual observations.
https://openreview.net/forum?id=3abgRKnK1W
Compressor summary: The authors propose a new learning framework for dissipative chaotic systems that targets the invariant measure and dynamics, resulting in better point-wise tracking and long-term statistical accuracy.
https://openreview.net/forum?id=3ZM8MXGFRA
Compressor summary: The paper presents Auto-Linear, a self-supervised method for subsurface imaging that achieves better performance, smaller model size, and stronger generalization than existing methods.
https://openreview.net/forum?id=3Z9CRr5srL
Compressor summary: The paper studies in-context learning (ICL) of neural language models using regular languages generated by random finite automata, and shows that Transformers outperform other sequence models by computing in-context n-gram statistics with specialized attention heads.
https://openreview.net/forum?id=3YG55Lbcnr
Compressor summary: The paper proposes a correlation clustering algorithm for dynamic vertex streams that approximates the optimal solution with low update time and performs well on real data.
https://openreview.net/forum?id=3XG69ZmfsB
Compressor summary: FreeBind is a method to enhance multimodal representation spaces by integrating knowledge from extra expert spaces using "space bonds", leading to improved performance on audio-image-text tasks.
https://openreview.net/forum?id=3WCvnkHnxV
Compressor summary: PrE-Text generates differentially private synthetic text data, enabling more efficient and privacy-friendly training of small models and finetuning large language models on user devices.
https://openreview.net/forum?id=3VnSgdget6
Compressor summary: The authors propose a novel method to combine multiple feature attribution techniques to improve the quality of explanations for opaque machine learning models, achieving better robustness and faithfulness to the model behavior.
https://openreview.net/forum?id=3Tzdpjc59k
Compressor summary: The paper proposes new dueling bandit models and algorithms for minimizing Borda regret in recommendation systems and ranking, achieving optimal regret bounds.
https://openreview.net/forum?id=3RXAiU7sss
Compressor summary: The paper advocates for publishing "negative" results in machine learning research to improve the scientific output and address issues caused by focusing solely on predictive performance.
https://openreview.net/forum?id=3QM5SWfeov
Compressor summary: Our novel meta-learning BO approach learns query utility, models task uncertainty, and adapts robustly to new tasks without relying on surrogate models.
https://openreview.net/forum?id=3Pq6uI1MTE
Compressor summary: The paper proposes a novel differentiable combinatorial scheduling framework that uses Gumbel-Softmax sampling to handle resource-constrained scheduling problems efficiently and effectively, outperforming existing solvers.
https://openreview.net/forum?id=3MfvxH3Gia
Compressor summary: The authors propose a novel multimodal survival method that compresses gigapixel histology images and transcriptomic profiles using morphological and pathway prototypes, enabling more efficient and interpretable patient prognostication and stratification.
https://openreview.net/forum?id=3McL91pE6x
Compressor summary: The paper introduces Logit-Smoothed ECE, a continuous and easy-to-estimate metric for measuring calibration, and compares it with existing methods like binned ECE on image classification models.
https://openreview.net/forum?id=3MW8GKNyzI
Compressor summary: Chatbot Arena is an open platform for evaluating large language models based on human preferences using pairwise comparisons and crowdsourcing, with a strong foundation of credibility and wide recognition.
https://openreview.net/forum?id=3MIuPRJYwf
Compressor summary: The paper presents a method to identify neural network weights using imitation and clustering of overparameterised networks with different activation functions.
https://openreview.net/forum?id=3KxPo62PYn
Compressor summary: The paper proposes adaptive models that can change their shape during training to reduce the required compute and outperform static models.
https://openreview.net/forum?id=3KMMPxrAk5
Compressor summary: The paper proposes a method to identify local causal structures from observational data that can handle latent variables, using m-separation and V-structures, with theoretical consistency results and experimental validation.
https://openreview.net/forum?id=3JhmHCVPa8
Compressor summary: The text proposes Merlin, a novel goal-conditioned reinforcement learning method based on denoising diffusion models, which improves performance and efficiency in offline goal-reaching tasks.
https://openreview.net/forum?id=3FeYlKIPr3
Compressor summary: Spiking neural networks with synaptic delay and temporal coding can perform human-like graph reasoning efficiently and energy-savingly.
https://openreview.net/forum?id=3FKEtlX4aM
Compressor summary: The paper proposes a hybrid approach combining PDE-based and deep learning models to improve inpainting of optical flow fields, achieving better performance than existing methods.
https://openreview.net/forum?id=3FBO41d4T2
Compressor summary: The paper introduces a new bound for the expectation of softplus function that improves variational logistic regression and Gaussian process classification by being tighter, faster and not requiring extending the variational family or adding extra parameters.
https://openreview.net/forum?id=3Cp042s1Nc
Compressor summary: The paper proposes a new rating system for large language models using a competitive format and a real-world questions benchmark, addressing issues with existing evaluation methods like MCQA.
https://openreview.net/forum?id=3B6vmW2L80
Compressor summary: The paper proposes a new model-free distributionally robust Q-learning algorithm (DRQ) that learns optimal policies from single trajectories with asymptotic convergence guarantees, achieving better robustness and sample complexity than existing methods.
https://openreview.net/forum?id=3AuoStfUIH
Compressor summary: The paper introduces a benchmark for offline multi-objective optimization (MOO) and analyzes how existing methods can be adapted to this challenging problem, with the goal of advancing the field.
https://openreview.net/forum?id=39UqOkTjFn
Compressor summary: The paper characterizes the utility of deterministic skills in sparse-reward environments with finite action spaces, showing that they are more beneficial for exploration than learning and that unexpressive skills may worsen performance.
https://openreview.net/forum?id=37xFIeYgE0
Compressor summary: The paper proposes distributional values, a new way to explain probabilistic machine learning models using cooperative game theory, by tracking changes in the model output for different actions or strategies.
https://openreview.net/forum?id=36rWa8zVkh
Compressor summary: The paper proposes SLIP, a single-loop algorithm for nonconvex bilevel optimization, which improves upon existing nested loop methods and achieves near-optimal complexity with no mean-square smoothness assumptions.
https://openreview.net/forum?id=36jWuAmGRC
Compressor summary: The paper proposes new projection-free algorithms for stochastic constrained multi-level optimization with improved complexities and applicability to various criteria and function types.
https://openreview.net/forum?id=35ahHydjXo
Compressor summary: The paper introduces Q-score matching, a new off-policy reinforcement learning algorithm that leverages the score-based structure of diffusion models for better actor-critic settings and exploration in continuous domains.
https://openreview.net/forum?id=321GwKMtxO
Compressor summary: REMEDI is a novel method for estimating information theoretic quantities like differential entropy with improved accuracy on synthetic and natural data, and can be extended to Information Bottleneck and generative modeling tasks.
https://openreview.net/forum?id=30waYPIZUA
Compressor summary: The authors develop a theory to understand and improve deep transformer models, enabling them to perform better across various tasks and datasets.
https://openreview.net/forum?id=2zLt2Odckx
Compressor summary: TFL is a method that uses client relationships graph to train robust models against unseen data in distributed settings.
https://openreview.net/forum?id=2zI2scD2Iz
Compressor summary: The paper proposes using hybrid reinforcement learning to reduce unnecessary exploration in imitation learning with inverse reinforcement learning, leading to better sample efficiency.
https://openreview.net/forum?id=2xbkWiEuR1
Compressor summary: The paper introduces AgentOptimizer, a novel method to train language models as agents without modifying their weights, improving their performance on various tasks.
https://openreview.net/forum?id=2xLyc5TkFl
Compressor summary: This paper investigates conformal prediction's uncertainty in adversarially trained models and proposes a new adversarial training method to improve predictive uncertainty.
https://openreview.net/forum?id=2ulUrcOZ64
Compressor summary: The paper proposes Switched FM, a method that solves singularity problems in continuous-time generative models by switching neural ODEs, improving sampling efficiency and compatibility with advanced techniques.
https://openreview.net/forum?id=2rPoTgEmjV
Compressor summary: The paper compares theoretical aspects of two generative self-supervised learning paradigms, autoregressive and masked, and proposes new objectives to improve their performance in classification and content generation tasks.
https://openreview.net/forum?id=2pYTCy4GUV
Compressor summary: Skip-Tuning is a training-free method that improves diffusion probabilistic models' performance and image quality by adjusting skip connections in UNet architectures.
https://openreview.net/forum?id=2hidpjUPvV
Compressor summary: The paper proposes an algorithm that uses one-layer ReLU neural networks to solve stochastic bandit problems with near-optimal regret by exploiting their piecewise linear structure and transforming the problem into a linear bandit.
https://openreview.net/forum?id=2hWd4CVhXz
Compressor summary: The paper explores how sample average approximation (SAA) can solve convex stochastic programming problems with less computational complexity.
https://openreview.net/forum?id=2dlmcTXfcY
Compressor summary: The paper introduces kernel-based tools using ACMMD to evaluate and tune conditional sequence models in computational biology, such as ProteinMPNN.
https://openreview.net/forum?id=2dEH0u8w0b
Compressor summary: TopKD is a novel knowledge distillation method that transfers global topology information from larger to smaller networks, using persistence diagrams to capture geometric structures in latent spaces.
https://openreview.net/forum?id=2cXzNDe614
Compressor summary: The paper proposes a new neural network architecture called PDHG-Net that combines first-order methods and learning to optimize to solve large-scale linear programming problems faster than existing methods.
https://openreview.net/forum?id=2cEhQ4vtTf
Compressor summary: SGDM is a graph generative framework that improves network quality by correcting corruptions, inferring missing nodes and edges, and performing conditional generation tasks.
https://openreview.net/forum?id=2bUFIsg2f5
Compressor summary: The paper proposes a new white-box attack method for generating sparse, $l_0$-bounded adversarial perturbations and shows that adversarial training can improve the robustness of models against such attacks.
https://openreview.net/forum?id=2Yu5FWdzde
Compressor summary: Prompt sketching is a new way to communicate with large language models by providing intermediate instructions during text generation, leading to better results and more control over the process.
https://openreview.net/forum?id=2Y93PtAqCl
Compressor summary: The study proposes a method to improve visual prompt tuning by initializing prompts with downstream token prototypes and optimizing token construction, achieving better performance and adaptability for downstream tasks.
https://openreview.net/forum?id=2XkRIijUKw
Compressor summary: The method predicts a distribution's quantile online with retrospective guarantee of coverage and improved practical properties for stable distributions.
https://openreview.net/forum?id=2W3KUAaZgO
Compressor summary: The paper proposes Operator INR, an alternative to Implicit Neural Representations that uses integral transforms and convolutions for better performance in compression, synthesis, and data understanding tasks.
https://openreview.net/forum?id=2T00oYk54P
Compressor summary: The Myerson-Taylor interaction index is a new tool for explaining how graph neural networks work by considering both node importance and graph structure, outperforming existing methods in experiments.
https://openreview.net/forum?id=2Sl0lPF6ka
Compressor summary: The paper analyzes the convergence behavior of a classification model called Mixture-of-experts with softmax gating and proposes modified gating functions to improve its performance.
https://openreview.net/forum?id=2RQqg2Y7Y6
Compressor summary: The paper shows that two alignment methods are equivalent, introduces a new method that combines them, and tests it on a summarization task.
https://openreview.net/forum?id=2PVjIQdq7N
Compressor summary: This paper proposes an efficient algorithm to learn the behavior of unknown dynamical systems over multilayer networks using few training examples, and analyzes the model complexity.
https://openreview.net/forum?id=2P6GVfSrfZ
Compressor summary: The text proposes a method called AurA to reduce toxicity in large language models by adjusting neuron activation levels based on their ability to discriminate toxic sentences, achieving significant reduction in toxicity while preserving common-sense abilities.
https://openreview.net/forum?id=2NfpFwJfKu
Compressor summary: The paper introduces UniCorn, a new pre-training framework for molecular foundation models that combines three existing methods and achieves state-of-the-art performance on various molecular tasks.
https://openreview.net/forum?id=2NUGeV64y2
Compressor summary: The authors propose a new method to defend against adversarial attacks using diffusion models guided by contrastive loss, which significantly improves performance on various datasets and classifiers.
https://openreview.net/forum?id=2K87GFLYWz
Compressor summary: The study explores the causes and solutions for learning plateaus in Transformers' in-context learning, improving their performance with three strategies.
https://openreview.net/forum?id=2JYOxcGlRe
Compressor summary: The text discusses using a combination of Reinforcement Learning and MDP homomorphisms to design efficient experiments over complex systems with natural geometries.
https://openreview.net/forum?id=2Gr5wZR6uc
Compressor summary: The text describes a new method, SLIPS, for sampling from unnormalized target distributions using stochastic localization techniques.
https://openreview.net/forum?id=2FKzbEE24s
Compressor summary: The text proposes a new differentiable partial observable generalized linear model that improves variational inference for learning neural connectivities from spike trains.
https://openreview.net/forum?id=2FHWFG5ahw
Compressor summary: AHAC is a model-based reinforcement learning algorithm that adapts the simulation horizon to avoid stiff dynamics, achieving better performance and efficiency than model-free methods in continuous control tasks.
https://openreview.net/forum?id=2B2U5kkGUA
Compressor summary: The paper explores how Sharpness-Aware Minimization, which modifies model weights instead of input samples, can improve adversarial robustness without sacrificing clean accuracy.
https://openreview.net/forum?id=2Asakozn3Z
Compressor summary: HarmoDT is a novel offline multi-task reinforcement learning method that uses meta-learning to find an optimal harmony subspace of parameters for each task, enhancing the performance of a unified policy.
https://openreview.net/forum?id=28SEr5iFyT
Compressor summary: The paper proposes a new Monte Carlo method, SHCV, that uses spherical harmonics as control variates to approximate the Sliced-Wasserstein distance with improved convergence rate and theoretical properties.
https://openreview.net/forum?id=283cGgWfM2
Compressor summary: ESM-AA is a novel approach for protein language modeling that can handle both atom and residue levels, improving performance in protein-molecule tasks.
https://openreview.net/forum?id=24zMewdzyJ
Compressor summary: The paper presents a new policy optimization method that combines risk-neutral algorithms with predicted CVaR contributions using reweighting.
https://openreview.net/forum?id=23tMOWscus
Compressor summary: The paper introduces new methods for estimating the feasible reward set of an expert agent from offline datasets in inverse reinforcement learning, considering the limitations of the data coverage.
https://openreview.net/forum?id=1zFkjbTgwC
Compressor summary: Quasi-Givens Orthogonal Fine-Tuning (qGOFT) is a method that improves parameter efficiency and downstream adaptation by using Givens rotations for orthogonal transformations in the parameter space.
https://openreview.net/forum?id=1ySQI9LE4w
Compressor summary: The paper proves that the stochastic proximal gradient method with Polyak momentum works well for non-convex problems regardless of batch size and shows its benefits in composite optimization settings.
https://openreview.net/forum?id=1xKgDANODx
Compressor summary: ReDream uses retrieval to improve text-to-3D generation by incorporating 3D geometry and adapting the diffusion model's prior.
https://openreview.net/forum?id=1wzdf6NjHd
Compressor summary: The text describes a new framework that trains deep neural networks with soft pseudo-labels using a meta-network-parameterized objective function, which improves performance and adapts to different tasks.
https://openreview.net/forum?id=1vGN3CSxVs
Compressor summary: EquiPocket is an E(3)-equivariant Graph Neural Network that predicts binding sites of target proteins more accurately than existing deep-learning methods by addressing their limitations and using a dense attention output layer.
https://openreview.net/forum?id=1v1oFF3aw0
Compressor summary: The text introduces a method for measuring uncertainty in predictive models when there are changes in the covariate and conditional relationships between the outcome and covariates.
https://openreview.net/forum?id=1tRLxQzdep
Compressor summary: The text introduces Pruner-Zero, an automatic framework for searching symbolic pruning metrics using genetic programming, which achieves better performance than existing post-training pruning methods for Large Language Models without retraining.
https://openreview.net/forum?id=1sesUtOIH5
Compressor summary: The paper proposes DecisionNCE, a universal objective that learns multimodal representations from image sequences and language instructions for autonomous robots, improving task progressions, temporal consistency, and instruction grounding.
https://openreview.net/forum?id=1sRuv4cnuZ
Compressor summary: The paper proposes a multi-track graph convolutional network that prevents heterophilic mixing and improves performance on several graph datasets by separating messages according to their category semantics.
https://openreview.net/forum?id=1puvYh729M
Compressor summary: The paper proposes ACE, a causality-aware actor-critic method with an entropy term and dormancy reset to improve exploration and performance in continuous control tasks.
https://openreview.net/forum?id=1pj0Sk8GfP
Compressor summary: The text describes how DDPMs can generate images in new regions of the data distribution, such as slightly smiling faces, by combining latent factors learned from separate subsets of the data.
https://openreview.net/forum?id=1oU4FKpVx5
Compressor summary: The paper proposes a pruning technique for MoE models that reduces model size and computation while preserving test accuracy.
https://openreview.net/forum?id=1nT6uc3HdY
Compressor summary: The text introduces pro-active DP, an optimization framework for DP-SGD that allows selecting parameters based on a fixed privacy budget and maximizing utility (test accuracy).
https://openreview.net/forum?id=1n3aC5rvdE
Compressor summary: Key points: - The paper proposes a new method for approximating LLA using a variational sparse GP - The method retains the DNN output as the predictive mean and allows for efficient stochastic optimization - The method outperforms existing efficient variants of LLA in terms of quality and computational time Summary: The paper introduces a novel sparse GP approach to approximate LLA for uncertainty estimation on DNNs, which preserves the DNN output and achieves better performance and efficiency than existing methods.
https://openreview.net/forum?id=1mf1ISuyS3
Compressor summary: The paper proposes techniques to extend certified unlearning methods to nonconvex deep neural networks, improving efficiency and addressing practical scenarios like sequential unlearning requests.
https://openreview.net/forum?id=1lDAGDe0UR
Compressor summary: CATS is a method that uses auxiliary time series generated from original time series to capture inter-series relationships for improved multivariate time series forecasting with less complexity and parameters.
https://openreview.net/forum?id=1khG2xf1yt
Compressor summary: The paper introduces νPI, a constrained optimization algorithm for neural networks that uses PI controllers to stabilize Lagrange multiplier updates and improve generalization.
https://openreview.net/forum?id=1jHiq640y1
Compressor summary: The paper proposes specialized solvers that improve Bayesian flow networks' sampling quality and speed by connecting them with diffusion models through stochastic differential equations.
https://openreview.net/forum?id=1dtYo5ywXZ
Compressor summary: The text explains how sparseness is a key factor in the success of MLP-based architectures like MLP-Mixer, and how they relate to sparse parameterization and Monarch matrices.
https://openreview.net/forum?id=1bJLl4fY6i
Compressor summary: The paper introduces an adjoint-equivariant neural network that works with data from any finite-dimensional semi-simple Lie algebra and demonstrates its effectiveness on different tasks.
https://openreview.net/forum?id=1ZJLNLZIpk
Compressor summary: The paper investigates local learning optimization for neural networks, proposes a gradient reconciliation strategy between neighboring modules, and shows improved performance and memory efficiency on ImageNet using CNN and Transformer architectures.
https://openreview.net/forum?id=1YsQI04KaN
Compressor summary: AbX is a new score-based diffusion generative model that uses evolutionary, physical, and geometric constraints to improve antibody design, outperforming other methods in accuracy and binding affinity.
https://openreview.net/forum?id=1YMjzz2g81
Compressor summary: The paper proposes SPABA, a method for bilevel optimization with optimal sample complexity, and shows its advantages over other stochastic gradient estimators in theory and practice.
https://openreview.net/forum?id=1YDeZU8Lt5
Compressor summary: The authors train and release OpenMoE, a series of open-source decoder-only MoE LLMs, analyze their routing mechanisms, and propose strategies to improve them for future LLM development.
https://openreview.net/forum?id=1WWpIEFdlk
Compressor summary: This paper proposes a diffusion-based image compression method that uses an end-to-end decoder to improve perceptual quality and guarantee distortion, by analyzing and improving the score function approximation in the diffusion model.
https://openreview.net/forum?id=1V50J0emll
Compressor summary: Physics-informed machine learning integrates observational data and physics models using Gaussian processes, and can benefit from Lie symmetry constraints to improve performance in forward and inverse problems.
https://openreview.net/forum?id=1SiEfsCecd
Compressor summary: PURE is a method to defend against backdoored language models by pruning and normalizing attention weights, reducing the attack success rate without harming clean text performance.
https://openreview.net/forum?id=1RZKuvqYCR
Compressor summary: The paper introduces TDPO, a novel method for fine-tuning LLMs at the token level, using forward KL divergence constraints and the Bradley-Terry model, which improves alignment with human preferences and generation diversity.
https://openreview.net/forum?id=1QmFKwVwwI
Compressor summary: The paper studies how to balance CATE estimation accuracy, patient outcome improvement (regret), and data privacy in adaptive experiments using contextual bandits and proposes Pareto optimal algorithms with differential privacy and asymptotic normality.
https://openreview.net/forum?id=1PMkV6oKw3
Compressor summary: The paper proposes Implicit Neural Teaching (INT), a method that improves the learning of implicit neural representations (INR) by treating it as a nonparametric teaching problem, which leads to faster convergence and reduced training time.
https://openreview.net/forum?id=1OsRSrkFWl
Compressor summary: The paper introduces two models for fair resource allocation in online settings, one based on external attributes like demand and the other based on internal traits like demographics, and proposes optimal policies for each model using different equity metrics.
https://openreview.net/forum?id=1NdN7eXyb4
Compressor summary: EAGLE is a speculative sampling framework that improves efficiency and preserves quality in LLMs by predicting second-to-top-layer features with advanced token sequences.
https://openreview.net/forum?id=1N7pjXKkx8
Compressor summary: The paper investigates how discrepancies between textual prompts used by protectors and exploiters affect existing defenses for privacy in few-shot fine-tuning of Latent Diffusion Models, and proposes a new method called Prompt-Independent Defense (PID) to safeguard privacy.
https://openreview.net/forum?id=1KemC8DNa0
Compressor summary: PruneX is a novel data-driven logic synthesis heuristic that reduces ineffective transformations by learning domain-invariant representations based on transformation-invariant domain knowledge and improves the efficiency of existing heuristics.
https://openreview.net/forum?id=1JgCpZS17T
Compressor summary: The paper proposes an algorithm to find changes in high-dimensional linear regression using a message passing method and shows its effectiveness on synthetic data and images.
https://openreview.net/forum?id=1IZLOPxtfK
Compressor summary: INViT is a new deep reinforcement learning architecture for solving routing problems that improves generalizability by using nested designs and invariant views, along with modified policy gradient and data augmentations.
https://openreview.net/forum?id=1HDrfUahXv
Compressor summary: The paper studies how well shallow autoencoders capture sparse data structure, showing that gradient descent ignores sparsity unless the data is very sparse, and proposes ways to improve compression for sparse data using denoising and multi-layer decoding.
https://openreview.net/forum?id=1Fs1LvjYQW
Compressor summary: The paper introduces MLAgentBench to test whether language model-driven agents can perform machine learning experimentation effectively and finds that Claude v3 Opus is the best performing agent.
https://openreview.net/forum?id=1DyruVvVaQ
Compressor summary: The paper introduces a new algorithmic framework for online linear programming that allows first-order methods to achieve better regret than previous approaches.
https://openreview.net/forum?id=1AAlMSo7Js
Compressor summary: DARE is a novel domain incremental learning method that reduces representation drift and catastrophic forgetting by gradually adapting new task representations to previous tasks' feature space and integrating task boundaries.
https://openreview.net/forum?id=18rzx2PXKm
Compressor summary: The paper introduces MOAC, an actor-critic algorithm for multi-objective reinforcement learning (MORL) that finds trade-offs among conflicting rewards and provides theoretical analysis of its convergence and sample complexity.
https://openreview.net/forum?id=18f6iPn0zq
Compressor summary: The paper studies layer normalization (LN), a common deep learning technique, focusing on its nonlinearity and representation capacity, and shows how to design neural architectures using LN to improve classification performance.
https://openreview.net/forum?id=181hXof7ho
Compressor summary: NeWRF is a deep-learning-based framework that uses Neural Radiance Fields to predict wireless channels, reducing the cost and effort of site surveys in wireless network deployments.
https://openreview.net/forum?id=17ZwoHl65h
Compressor summary: PlanDQ is a hierarchical planner for offline RL that combines a diffusion-based high-level planner with a Q-learning based low-level policy to handle sparse-reward and long-horizon tasks.
https://openreview.net/forum?id=15MpDbv3IQ
Compressor summary: The algorithm estimates CDF values under LDP using a connection with the current status problem and tools for constrained isotonic estimation based on binary queries, achieving error bounds that improve with more grids.
https://openreview.net/forum?id=126SR50BEL
Compressor summary: The text introduces ACTIN, a novel framework for estimating counterfactuals that uses adversarial methods to balance representations and temporal integration to capture long-range dependencies and interactions, achieving state-of-the-art performance with simple base models.
https://openreview.net/forum?id=10hu2D3hAg
Compressor summary: The paper proposes SIFT, a gradient-based sparse fine-tuning algorithm for pre-trained models, which tightens the generalization error bound by shifting the prior distribution and leverages oscillations in the loss landscape and quasi-sparsity in gradient distribution.
https://openreview.net/forum?id=0zbxwvJqwf
Compressor summary: LatProtRL is a new method to optimize protein functions using a large language model and reinforcement learning, which could benefit various industries.
https://openreview.net/forum?id=0xmfExPqFf
Compressor summary: The paper introduces a general framework for risk-sensitive distributional reinforcement learning with static risk measures and different function approximation methods, and presents two novel meta-algorithms with improved regret bounds.
https://openreview.net/forum?id=0wso32h0jc
Compressor summary: The paper proposes a hybrid method that combines deep learning and kernel conditional mean embeddings to address scalability and expressiveness challenges in representing conditional distributions, and shows its effectiveness in density estimation and distributional reinforcement learning tasks.
https://openreview.net/forum?id=0vozy8vstt
Compressor summary: The paper develops contextual algorithms for bandits with uninformed feedback graphs by reducing online regression over losses and graphs, and shows that using log loss for graph learning is crucial for good performance.
https://openreview.net/forum?id=0urN0PnNDj
Compressor summary: PEARL is a novel RL method that learns policies from cross-task preference transfer without human labels, using optimal transport to align trajectories and Gaussian distributions to model reward uncertainty.
https://openreview.net/forum?id=0uUHfhXdnH
Compressor summary: The paper introduces DiJiang, a method to linearize Transformers in the frequency domain using kernels based on Discrete Cosine Transform and weighted Quasi-Monte Carlo sampling, achieving comparable performance with reduced training costs and faster inference speeds.
https://openreview.net/forum?id=0tuwdgBiSN
Compressor summary: The paper proposes a framework to study the impact of spurious features on neural network learning dynamics and reveals several phenomena about core and spurious feature learning, validating existing debiasing techniques and highlighting their limitations.
https://openreview.net/forum?id=0tYrMtQyPT
Compressor summary: Log-NCDEs use a new method based on Log-ODEs to train neural differential equations that model real-world data better than existing approaches.
https://openreview.net/forum?id=0tPBk24xNj
Compressor summary: The paper studies reward poisoning attacks on Combinatorial Multi-armed Bandits (CMAB), and finds that the attackability of a CMAB instance depends on its properties and whether it is known or unknown to the adversary.
https://openreview.net/forum?id=0rV7VIrcjX
Compressor summary: The paper introduces Graph External Attention, a novel attention mechanism that leverages external node/edge units to capture inter-graph correlations, and proposes the Graph External Attention Enhanced Transformer (GEAET), which improves graph representation learning by integrating local and global information.
https://openreview.net/forum?id=0pSTzCnEmi
Compressor summary: The text introduces LA-NAMs, a Bayesian approach to neural additive models that provides uncertainty estimates, feature selection, and second-order interaction ranking for deep neural networks.
https://openreview.net/forum?id=0ofzEysK2D
Compressor summary: The authors propose a framework to classify AGI models by their performance, generality, and autonomy levels, and discuss how it can help compare, assess, and measure progress in AGI research.
https://openreview.net/forum?id=0ntak1BGBd
Compressor summary: ED-Copilot is an AI system that suggests laboratory tests and makes diagnoses faster and more accurately in emergency departments, potentially reducing crowding and improving patient outcomes.
https://openreview.net/forum?id=0nMzOmkBHC
Compressor summary: The paper proposes FedSC, a provable Federated Self-Supervised Learning algorithm that uses spectral contrastive objective, sharing correlation matrices, and deploying differential privacy protection to improve data representations and performance.
https://openreview.net/forum?id=0mklK4h0rX
Compressor summary: ExSASCA is a fast and accurate method to detect weaknesses in cryptographic algorithms using knowledge compilation and probabilistic circuits, improving on the existing SASCA attack by over 31%.
https://openreview.net/forum?id=0miAQ1qHiw
Compressor summary: The paper investigates structured variational families, a middle ground between mean-field and full-rank ones, that can improve the efficiency of black-box variational inference for hierarchical Bayesian models.
https://openreview.net/forum?id=0mYAK6Yhhm
Compressor summary: The authors propose a new computational method, CLIPZyme, to identify efficient catalysts among uncharacterized proteins by encoding and aligning enzyme structures and reactions.
https://openreview.net/forum?id=0ksNeD1SJT
Compressor summary: Key points: - The paper proposes a new perspective on parameterization and derives new theoretical results - The paper conducts extensive empirical investigation with many combinations of optimizers, parameterizations, assumptions, learning rates, and model sizes - The paper finds that prior work's assumptions often exclude the best learning rate scaling prescription - The paper introduces Adam-atan2, a new scale-invariant version of Adam that eliminates the epsilon hyperparameter Summary: The paper presents a novel parameterization perspective, investigates many algorithmic and architectural details, reveals flaws in prior work's assumptions, and proposes Adam-atan2, a numerically stable optimizer.
https://openreview.net/forum?id=0jpbpFia8m
Compressor summary: SqueezeLLM is a post-training quantization framework that compresses large language models with ultra-low precision, improving performance and reducing memory requirements for inference.
https://openreview.net/forum?id=0j28mmQ023
Compressor summary: The paper proposes a data acquisition framework that improves machine learning model performance on shifted data by refining the preparation of training data from various domains.
https://openreview.net/forum?id=0iXp5P77ho
Compressor summary: Tripod is a neural network autoencoder with three complementary inductive biases that enable disentangled representation learning, achieving state-of-the-art results on image disentanglement benchmarks.
https://openreview.net/forum?id=0hbeZQm1Se
Compressor summary: DCG is a novel algorithm framework for solving sequential adversarial team games with asymmetric information by transforming coordinated best responses into TB-DAG form, which converges exponentially faster and is more scalable than CG algorithms.
https://openreview.net/forum?id=0f4u3Wg9zT
Compressor summary: The authors propose HyRA, a method to interpret deep neural networks' behavior in image super-resolution using signal processing theories, and introduce FSDS, a metric to measure high-frequency information injection.
https://openreview.net/forum?id=0e8SEDSpNT
Compressor summary: Key points: - Dynamic convolution learns a linear mixture of static kernels with input-dependent attentions, but is not parameter efficient - KernelWarehouse proposes a more general form of dynamic convolution that exploits convolutional parameter dependencies within and across layers - KernelWarehouse improves accuracy on various ConvNet architectures and Vision Transformers, and reduces model size in some cases Summary: KernelWarehouse is a novel dynamic convolution method that leverages convolutional parameter dependencies to achieve better performance and efficiency than normal convolution.
https://openreview.net/forum?id=0bmXrtTDUu
Compressor summary: The paper modifies the Chinchilla scaling laws to account for inference costs and suggests training smaller and longer LLMs with optimal token/parameter ratios for large inference demand.
https://openreview.net/forum?id=0bGsVoumFL
Compressor summary: The paper analyzes why entropy minimization (EM) works initially but fails eventually for classification model adaptation and proposes a method to estimate a model's accuracy without labels using EM.
https://openreview.net/forum?id=0b7txvPYlr
Compressor summary: The paper introduces a statistical framework to control the accuracy gap between full and early-time classification by applying a calibrated stopping rule to any sequential classifier.
https://openreview.net/forum?id=0ZTuy5CrL7
Compressor summary: TVE is a tool that explains how various vision models work on different tasks without needing extra training data.
https://openreview.net/forum?id=0ZFWfeVsaD
Compressor summary: The paper proposes a method to build and reuse adapters for large language models on new tasks using model-based clustering and zero-shot routing, achieving better generalization than existing approaches.
https://openreview.net/forum?id=0XDO74NlOd
Compressor summary: The paper studies how different graph generative models balance representation power and diversity, introduces new models for each level of complexity, and evaluates them on real datasets.
https://openreview.net/forum?id=0THUA66D8Z
Compressor summary: The paper proposes CCNet, a method to improve the calibration of confidence scores in 2D human pose estimation by learning network-specific adjustments.
https://openreview.net/forum?id=0SrNCSklZx
Compressor summary: SLOG is a novel spectral graph neural network that overcomes the limitations of existing spectral GNNs by using a real-valued filter and combining subgraph sampling with signal processing for large-scale inductive node classification.
https://openreview.net/forum?id=0P3kaNluGj
Compressor summary: The paper proposes a neuro-symbolic framework that learns structured states and symbolic policies using a vision foundation model, and generates textual explanations with GPT-4 to improve interpretability of decision-making.
https://openreview.net/forum?id=0NphYCmgua
Compressor summary: This paper proposes self-rewarding language models that improve their own training by using a large language model as a judge and shows that they outperform existing systems on a leaderboard.
https://openreview.net/forum?id=0NdU4y9dWC
Compressor summary: DISGEN improves the size generalization of graph neural networks by disentangling size factors from graph representations using augmentations and a decoupling loss.
https://openreview.net/forum?id=0NacraIYrA
Compressor summary: The paper proposes an efficient method to approximate a hard financial risk management problem using indicator functions and algorithms that adjust the asset pool size.
https://openreview.net/forum?id=0M2tNui8jX
Compressor summary: Global Reinforcement Learning (GRL) introduces rewards defined globally over trajectories, capturing interactions among states that classic RL cannot model, and proposes a novel algorithmic scheme to solve GRL problems efficiently.
https://openreview.net/forum?id=0LBNdbmQCM
Compressor summary: The text describes a novel pre-training purification method for defending against unlearnable examples in image classification using disentangled variational autoencoders and a two-stage purification approach.
https://openreview.net/forum?id=0JXGusc7E2
Compressor summary: SeemoRe is a novel image super-resolution model that efficiently combines experts at different levels to reconstruct high-resolution images from low-resolution inputs.
https://openreview.net/forum?id=0JV5WpLQgv
Compressor summary: PointMC is a novel point cloud registration framework that uses maximal cliques and local spatial consistency to accurately estimate multiple rigid transformations for overlapping instances.
https://openreview.net/forum?id=0IDaPnY5d5
Compressor summary: The paper proposes a new reinforcement learning method that uses auxiliary tasks with short delays to improve learning with long delays, achieving better sample efficiency and policy performance than existing methods.
https://openreview.net/forum?id=0HUInAsdoo
Compressor summary: Key points: - The paper proposes OxyGenerator, a deep learning model for global ocean deoxygenation reconstruction from 1920 to 2023. - The model uses zoning-varying graph message-passing and inductive bias to capture complex correlations and uncertainty in DO variations. - OxyGenerator outperforms CMIP6 numerical simulations by reducing MAPE by 38.77%. Summary: OxyGenerator is a deep learning model that reconstructs global ocean deoxygenation from 1920 to 2023 using zoning-varying graph message-passing and inductive bias, outperforming CMIP6 simulations.
https://openreview.net/forum?id=0GC0NG6Orr
Compressor summary: GST is a fast and flexible optimal transport method for graph metric spaces, improving on Sobolev transport and Orlicz-Wasserstein.
https://openreview.net/forum?id=0FWPKHMCSc
Compressor summary: The paper introduces Dataset Distillation with Domain Shift (D3S), a scalable method to summarize large datasets by reframing the problem as a domain shift issue.
https://openreview.net/forum?id=0AZAjkXhit
Compressor summary: The authors propose a simple and effective method of selecting long instructions as the basis for fine-tuning LLMs, which outperforms state-of-the-art methods and achieves competitive results on various benchmarks.
https://openreview.net/forum?id=09Robz3Ppy
Compressor summary: The paper proposes a new score-based algorithm for learning causal structures from missing data using optimal transport, which outperforms existing methods in simulations and real data settings.
https://openreview.net/forum?id=07fSWltF6M
Compressor summary: ProtoGate is a neural model that selects features on high-dimensional low-sample-size data by balancing global and local approaches and using prototype-based predictions to avoid co-adaptation problem.
https://openreview.net/forum?id=07f24ya6eX
Compressor summary: The paper introduces 2RA Q-learning, a new method to control estimation bias in Q-learning, and shows its improved performance over existing methods.
https://openreview.net/forum?id=051jaf8MQy
Compressor summary: PIVOT is a visual prompting method that helps vision language models perform tasks like robotic control by iteratively refining proposals based on image annotations.
https://openreview.net/forum?id=04Fx1u2BUD
Compressor summary: SSL4Q is a new semi-supervised learning method for quantum state classification that works well with limited labeled data and requires fewer resources than traditional methods.
https://openreview.net/forum?id=01ahsMovBx
Compressor summary: MetaFormer is a new framework that uses self-attention to enhance pre-trained vision transformers for few-shot image classification by embedding sample relationships and consolidating task patterns.
https://openreview.net/forum?id=01M0N8VgfB
Compressor summary: The paper proposes a method to improve server-side training simulations for federated learning by partitioning centralized data based on the statistical heterogeneity of the true clients.