AI News

A Missing Latent, Not a Missing Simulator: Radius-Augmented Inference for Real JWST Retrieval (arxiv.org)

Causal discovery identifies pathways linking physical activity to dementia risk in the UK BioBank (arxiv.org)

CORE: COverage CAlibration and Evicted-Mass REdistribution for KV Cache (arxiv.org)

RxnOptBench: Benchmarking LLMs for Reaction-Condition Optimization in Organic Methodology (arxiv.org)

Toward Controlling Biology with Language:Offline Learning of Prompt-Conditioned Interventions for Cells, Organoids, and Biobots (arxiv.org)

Revisiting Visual Representation Enhancement of VLMs via Kernel Canonical Correlation Analysis (arxiv.org)

Learning Query Encoders Can Be Hard Even When Vector Retrieval Is Geometrically Easy (arxiv.org)

Localized Conformal Safety Monitoring with Vision-Language Models for Autonomous Driving (arxiv.org)

RMCW: A Deletion-Robust Watermark Based on Reed--Muller Codes for Language Models (arxiv.org)

Autoregressive Differentiable Method for Integer Programming (arxiv.org)

FastOPD: On-Policy Distillation for Lightweight VLA Deployment (arxiv.org)

Turnover-Orthogonal Credit Assignment for Open-Team Multi-Agent Reinforcement Learning (arxiv.org)

Permutation Robustness Is Not Enough: Action Collapse in Multi-Agent Transformer Policies (arxiv.org)

Discriminating Fixture Coverage in Agent-Infrastructure Verification Suites (arxiv.org)

LEAP: Learning Efficient Action Proposals For LLM Agents (arxiv.org)

Test-time Calibration Learning for Large Language Model Reasoning (arxiv.org)

MuonIO: Principled Norm-Aware Descent for Embedding Tables and Language Model Heads (arxiv.org)

Self-Supervised Scaling of Terminal Environments for Scientific Domains (arxiv.org)

Reward Inflation: A Healthy Stimulus for Reinforcement Learning (arxiv.org)

TPBench: A Turning-Point Benchmark for Dialogue Compression (arxiv.org)

Prospective Hindsight: Self-Calibrating Reinforcement Learning via Prediction-Reality Gaps (arxiv.org)

Correcting Guided Diffusion Trajectories with Spectral Alignment (arxiv.org)

When History Fails to Become Experience: Action Calibration in Language Agents (arxiv.org)

Inner Momentum for Differentially Private Muon (arxiv.org)

VisAudit: Evaluating Multimodal Agents for Visual Diagnosis and Repair (arxiv.org)

Trained Agentic Context Management (arxiv.org)

Efficient Neural Field Learning via Adaptive Coverage and Focused Sampling (arxiv.org)

Prompted to Discriminate: Generalizing Malicious-Input Probes in the Wild (arxiv.org)

The AI Theorist reveals excitonic structure in $\alpha$-RuCl$_3$ (arxiv.org)

Geometry-Aware Time Reparameterization for Flow-Map Distillation (arxiv.org)

Improving Atomic-Fact Recall via Focused Views in Unstructured Knowledge Editing (arxiv.org)

OPD Before RL: Warm-Starting Rubric-Based RL with On-Policy Distillation (arxiv.org)

Adaptive Spectral-Koopman Dynamics Modeling for Temporal Domain Generalization (arxiv.org)

Distributionally Robust Survival Models under Subpopulation Shift and Outlier Contamination (arxiv.org)

CRISP: A Framework for Clause-Reconstructed Interpretable NeuroSymbolic Propositions (arxiv.org)

Are you Synthesizing or Recalling? Evaluating LLMs on Algorithmic Code Retrieval (arxiv.org)

A Generative Model of Complex Networks Using Graphons and Neural Inverse Operators (arxiv.org)

Bandits via Additive Quantized Representations (arxiv.org)

AI-driven Thermal-aware Data Center Capacity Planning (arxiv.org)

AgentTrap: Stateful Feedback Deception against Autonomous Penetration Testing Agents (arxiv.org)

ConvoDrift: A Multi-Turn Conversational Dataset for Modeling Stylistic Tone Evolution (arxiv.org)

Query-aware routing for Cross-lingual performance gains in Encoders (arxiv.org)

DyRA: Dynamic Residual Approximation for Efficient Matrix Multiplication in DNNs (arxiv.org)

Closing the Train-Test Gap in World Models for Gradient-Based Planning (arxiv.org)

A Composable AI-Accelerated Iterative Solver for 3D-IC Thermal Modeling (arxiv.org)

Capability Scaling-Down Laws for LLM Compression (arxiv.org)

Harnessing LLMs as Agents: What Does It Cost? (arxiv.org)

LiteEMG-FM: An Efficient and Deployable Foundation Model for Robust EMG Sensing (arxiv.org)

Recoverability Has a Law: The ERR Measure for Tool-Augmented Agents (arxiv.org)

Misinformation Without Triggers: From Factual Answers to Downstream Decisions (arxiv.org)