AI News

Local Support Learning (arxiv.org)

Are We Recovering Mechanisms? Objective-Level Recovery Gaps in Mechanistic Interpretability (arxiv.org)

Learn the Directions, Normalize the Gains: Post-Training Normalization for LoRA (arxiv.org)

Encoded but Disconnected: Decomposing Vision-Language Model Failures under a Patching Null (arxiv.org)

Foundations without Fundamentals: Zero-Shot Blind Spots in Time Series FMs (arxiv.org)

Distributionally Robust Schr\"odinger Bridge (arxiv.org)

On Evaluating Quantum Kernel Robustness for Low-Resource Cross-Corpus Audio Deepfake Detection (arxiv.org)

CARM: Cancellation-Aware Response Masking for LLM Reinforcement Learning (arxiv.org)

Relative Transitions, Not Absolute Destinations: A Transfer-and-Ground Framework for Target-Trajectory-Free Human Mobility Generation (arxiv.org)

On Language Drift during RLVR Post-Training (arxiv.org)

BranchIP: Learning Adaptive Equivariant Computation for Interatomic Potentials (arxiv.org)

Bellman Meets Lyapunov: Unsupervised Reinforcement Learning via Mastering Chaos (arxiv.org)

Universal interpolation for deep residual self-attention networks (arxiv.org)

Clifford Sheaf Neural Networks (arxiv.org)

Safety of Latent Communication in Multi-Agent Systems (arxiv.org)

CATCH: A Controllable Analysis Testbed for Reward Hacking in Coding RL (arxiv.org)

Bandits with Multiple Optimal Arms: Minimax Regret and Non-Adaptivity (arxiv.org)

Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI (arxiv.org)

Identifiability Guarantees for Drivers and Dynamics of Delayed Physical Systems (arxiv.org)

Online Generalized-Mean Welfare Maximization: Achieving Near-Optimal Regret from Samples (arxiv.org)

Trajectory Stitching for Solving Inverse Problems with Flow-Based Models (arxiv.org)

Rethinking Anonymity Claims in Synthetic Data Generation: A Model-Centric Privacy Attack Perspective (arxiv.org)

In Vino Veritas and Vulnerabilities: Examining LLM Safety via Drunk Language Inducement (arxiv.org)

VISPA: Pluralistic Alignment via Automatic Value Selection and Activation (arxiv.org)

In CEM, a World Model Is Also a Proposal Mechanism (arxiv.org)

Sharpen Before You Adapt: Data-Free Entry-State Sharpening for Test-Time Reinforcement Learning (arxiv.org)

WOMBAT: Whitebox Oracle for Molecular Benchmarking and Attribution Testing (arxiv.org)

Scalable, Transferable Meta-network for Data Selection Requires a Different Loss (and Why the Obvious Choice is Problematic) (arxiv.org)

Kolmogorov-Arnold Networks for Free-Boundary Partial Differential Equations (arxiv.org)

Lower Bounds for Stochastic First-Order Algorithms with Variance Reduction in Nonconvex--Concave Minimax Optimization (arxiv.org)

QATFactory: A Versatile, Deployment-Aligned Framework for Quantization-aware Training and Distillation of LLMs (arxiv.org)

Explainability of Complex AI Models with Correlation Impact Ratio (arxiv.org)

Poincar\'e Meets Bellman: Revisable Memory, Operational Quotients, and Evidence-Supported Learning in Changing Environments (arxiv.org)

Which LLM to pick? Online Active Model Selection for Large Language Models (arxiv.org)

Continual Reinforcement Learning with Neuroevolution (arxiv.org)

The hidden advantage of mask resampling: a theory of masked autoencoders (arxiv.org)

Optimal Momentum Methods for Stochastic Multilevel Compositional Optimization (arxiv.org)

Hob-VL: A Benchmark for Visually Grounded Boolean Reasoning (arxiv.org)

Managing Context and Communication in Distributed Agentic UAV Swarms (arxiv.org)

AURAL: Adaptive Latent Reasoning with Joint Chunk for Speech Language Models (arxiv.org)

Neither Black nor White: Balancing Semantic and Collaborative Signals with Graph-Informed Semantic IDs (GrIS) (arxiv.org)

Decision Titan: Test-Time Training for Long-Term Memory in Offline Reinforcement Learning (arxiv.org)

Sharpening Tax in Post-Training (arxiv.org)

OpenMTB-Audit: Exposing Over-Refusal and Clinical Expert Perspectives in LLM-Based Molecular Tumor Board Safety Evaluation (arxiv.org)

Completion Aware Guidance for World Action Models (arxiv.org)

Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Training (arxiv.org)

Prediction-powered Neural Architecture Search (arxiv.org)

EP-Flow: Disordered Crystal Structure Prediction without Site-Level Annotations (arxiv.org)

Model validation in machine learning: A scenario-based guide from hold-out splits to nested group cross-validation in biomedical and applied research (arxiv.org)

Reinforcement Learning to Accelerate Primal-Dual Hybrid Gradient for Linear Programming (arxiv.org)