AI News

Learning Topology-Aware Representations for Mechanics-Relevant TPMS Metamaterials (arxiv.org)

Diminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility (arxiv.org)

Designing Experiments for External Decision Makers (arxiv.org)

Robust Covariance Estimation for Entry-wise Contaminated and Partially Observed Data via Expectation Maximization Algorithm (arxiv.org)

Conditional Mean Flows: Few-Step Densities for Simulation-Based Inference (arxiv.org)

LCT: Losslessly Compressed Tensors with GPU friendly execution for Training (arxiv.org)

When Is Muon Optimal? Radial Fragility and Minimax Geometry (arxiv.org)

Two-Stage Score Estimation for Group-Specific Diffusion Models (arxiv.org)

Stochastic Gradient Descent: A Nonlinear Time Series Perspective (arxiv.org)

Complete or Sparse: A Tale of Two Identifiabilities (arxiv.org)

On the Stability and Convergence of Flow Map Training (arxiv.org)

In-Context Learning Before Specialisation: Theory of Learning in a Solvable Topic Model (arxiv.org)

Classical methods for modern applications: synthetic tabular data generation using lightweight discrete response continuous latent variable model (arxiv.org)

Sharp and Stable Label-Free Calibration Certificates from Logical Coherence (arxiv.org)

Two Sample Sizes in Test-Time LLM Ensembling: Buffered ERM and Minimax Theory (arxiv.org)

Controlling the False Discovery Rate via Model-Agnostic Importance Measures (arxiv.org)

A One-Sided McDiarmid-Type Inequality for Heavy Tails via Reverse KL Tensorization (arxiv.org)

StochLOB: Heavy-Tailed Score Matching for Robust Limit Order Book Trend Prediction (arxiv.org)

On the Last-Iterate Convergence of Clipped Gradient Methods (arxiv.org)

Failure Is the Mother of Success: Recovering Supervision with Grounding Hindsight (arxiv.org)

Dimension-Free PAC-Bayes Generalization for Deep Neural Networks via Dirichlet Processes (arxiv.org)

Score Mismatch and Set-Constrained Sampling in Zero-Shot Conditional Discrete Diffusion (arxiv.org)

Efficient Agnostic Learning of Univariate Densities Under Local Differential Privacy (arxiv.org)

Acceleration and its Limits in Gaussian Variational Inference (arxiv.org)

Halve-Then-Greedy: Adaptive Sampling for Posterior Success with Near-Optimal Instance Dependence (arxiv.org)

Flexible Doubly Robust Proximal Causal Inference for Censored Outcomes (arxiv.org)

Time-partitioning Gaussian Process Mixture Model (arxiv.org)

Latent Annealed Posterior Sampling with a Variance-Preserving Bridge for Sparsely Observed Random Fields (arxiv.org)

A polylogarithmic bound for list replicability (arxiv.org)

PAC-Causal Learnability: From Identification to Finite-Sample Guarantees (arxiv.org)

Distributionally Robust Regularization of Sparse Integer Programming Trained Learning Models (arxiv.org)

Explore Once, Evaluate Many: Multiple Policy Evaluation for Stochastic Linear Bandits (arxiv.org)

Query-Efficient Rejection Sampling for Parallel Masked Diffusion (arxiv.org)

Why Shallow Neural Networks Learn Sinusoidal Components First: An Orthogonal Polynomial Dynamics Framework (arxiv.org)

Market Incentives for AI Safety Investment (arxiv.org)

Regret Lower Bound for Online Learning of Quantum States with Logarithmic Loss, with Applications to Quantum State Tomography and Confidence Sequences for Matrix Means (arxiv.org)

Spectral Indistinguishability and Learning-Curve Identifiability in Quadratic Networks (arxiv.org)

SurvICL: a Foundation Model for in-context Survival Analysis (arxiv.org)

Learning Changes the World and Then What Was Learned has Changed: Training and Deploying Neural Networks in the Control Loop (arxiv.org)

Rethinking Successive Rejects Algorithms (arxiv.org)

The importance of tensor handling and implementation of physics-informed DeepONets (arxiv.org)

CoTS: Competitive Token Selection for Latent Reasoning in Large Language Models (arxiv.org)

Sparse Federated Echo State Networks for Time-Series Classification via Smooth-L1 Regularization (arxiv.org)

ShiftQAT: Quantization-Aware Training for the Numerical Contract of a Fixed-Function Edge CNN Accelerator (arxiv.org)

Rethinking Learning from Label Proportions via Moment Matching (arxiv.org)

Behavioral Compositionality of Sparse Autoencoder Representations (arxiv.org)

Learning to Read Out: Unembedding Dynamics in Language Model Pretraining (arxiv.org)

Rethinking Routing Exploration in Modern Mixture-of-Experts (arxiv.org)

Lean Graph Search for Formal Theorem Proving (arxiv.org)

Resisting Harmful Instructions through Training-Free Weight Editing (arxiv.org)