AI News

When Does Synthetic Relational Data Teach Models to Use Relations? Tracing Predictive Structure from Pretraining Data to Model Behavior (arxiv.org)

Learning Transferable Policies from Action-free Time Series Through Dynamical Embeddings (arxiv.org)

Learn Feasibility Once, Optimize All Objectives: Derivative-Free Diffusion Models for Chance-Constrained Programming (arxiv.org)

Smart Sensing for Safer Bridges: From Sensor Signals to AI-Driven Anomaly Detection (arxiv.org)

Light Entropic Optimal Transport on Riemannian Manifolds (arxiv.org)

LS-AR: Future-Predictive Latent Steering in Autoregressive LLMs (arxiv.org)

Exploring the Trade-Off Between Structured Pruning and Fault Tolerance in Deep Neural Networks for Space Applications (arxiv.org)

Coverage You Can Steer: Online Conformal Calibration for RL-Driven Hardware-Aware NAS (arxiv.org)

Page-EntroKV: Hardware-Aligned, Entropy-Weighted KV-Cache Eviction under Grouped-Query Attention (arxiv.org)

Sample complexity of variance-reduced policy gradient: weaker assumptions and lower bounds (arxiv.org)

ConfAL-WM: Confidence-Guided Active Learning for Action-Conditioned World Models (arxiv.org)

Predicting and Repairing Merge Collapse in Large Language Models (arxiv.org)

HyperFuse: Fast Self-Supervised Node Embeddings for Attributed Hypergraphs (arxiv.org)

Kernel Singular Value Decomposition with Extension to Multiple Data Sources (arxiv.org)

AdaStep: Adaptive Step Credit Weighting for Agentic Reinforcement Learning (arxiv.org)

The Neuro-Physical Inverter: A Modular Framework for Magnetotelluric Inversion Coupling Ensemble Conditioning with Residual Learning (arxiv.org)

Efficient Support Recovery of Mixtures of Sparse Linear Classifiers with Fewer Measurements (arxiv.org)

Cross-cohort TB classification using clinical data gathered in Uganda and South Africa (arxiv.org)

PaMIR: Open Benchmark of Public Credit-Default Datasets (arxiv.org)

Balancing Multimodal Learning via Functional Progress (arxiv.org)

What Does a ProcGen Generalization Gap Measure? Action Rules, Residual Entropy, and the Missing Random Floor (arxiv.org)

MaskCoFT: Masked Co-Adaptive Fine-Tuning for Memory-Efficient MoE Inference (arxiv.org)

MASCIT: A Mask-Aware State Space Classifier for Naturally Irregular Time Series (arxiv.org)

Embedded Bi-Temporal Building Damage Assessment for On-Board Data Reduction (arxiv.org)

ARGOS: Reinforcement Learning-Driven Multidimensional Elasticity for Service Orchestration in the Computing Continuum (arxiv.org)

Auditing Long-Term Memory Evaluation: Repeated Judging, Reader Variation, and Negative Controls (arxiv.org)

Science Is Falling Behind the Frontier: Foundation Model Adoption Across Half a Million Papers (arxiv.org)

From Solo to Social Learning: Characterizing Recursive Social Improvement in LLMs (arxiv.org)

CRAFT: Causal Responsibility and Failure Tracing in Medical Vision Language Models (arxiv.org)

Tactile Curiosity Drives Robot Interaction (arxiv.org)

Automated Feature Engineering, AutoML, and Decision-Focused Learning for Improved Energy Consumption Forecasting (arxiv.org)

From Migration to Calibration: Preserving Agent Capabilities across Models, Jurisdictions, and Scale (arxiv.org)

SafeCoEvo: Co-Evolving Safety Harnesses and Guards for LLM Agents at Test-Time (arxiv.org)

From Learner Behavior to Reusable Skills for Effective and Efficient Learner Simulation (arxiv.org)

WISE-ATTA: When to Ask for Labels in Budgeted Active Test-Time Adaptation (arxiv.org)

Learning to Route in Visual Space via Multi-Step Embedding Retrieval (arxiv.org)

Autoresearch in Mixed-Integer Linear and Nonlinear Programming (arxiv.org)

PTNO: Training Neural Operators with Noisy Monte Carlo Estimates for Particle Transport Problems (arxiv.org)

Cogentic: Multi-Agent Orchestration for Automated Proof Discovery (arxiv.org)

Rethinking Probability-Based Reinforcement Learning From Posterior Concentration (arxiv.org)

Causal Memory Policy: Making Memory Utility Identifiable by Intervening on Retrieval (arxiv.org)

Learning Low-Frequency Motion Control for Robust and Dynamic Robot Locomotion (arxiv.org)

ETHER: Aligning Emergent Communication for Hindsight Experience Replay (arxiv.org)

VIDiff: Translating Videos via Multi-Modal Instructions with Diffusion Models (arxiv.org)

Towards Unified Music Emotion Recognition across Dimensional and Categorical Models (arxiv.org)

Learning Skills from Historical Action Trajectories: Action Experience Dictionary for World Action Models (arxiv.org)

Screw Attention: Rigid-Body Algebra Inside a Transformer (arxiv.org)

Fold'EM: Direct atomic structure inference from Cryo-EM particles (arxiv.org)

Cross-Lingual Alignment for Decoder-Only Models using MoE Routers (arxiv.org)

Hybrid Reasoning Systems That Prioritize and Enhance Human Intelligence (arxiv.org)