AI News

Adaptive Replay Refinement: Allocating Test-Time Compute Between Revision and Resampling (arxiv.org)

Local Work, Global Progress: A Convergence Theory of AI Agent Collaboration (arxiv.org)

What Do RTL Repair Benchmarks Actually Measure? (arxiv.org)

TerraNostra-Bench: Benchmarking Spatial Downscaling and Temporal Rollout for Global 1-km Land-Use Reconstruction (arxiv.org)

Give the Model a Finger: Developmentally Aligned Evaluation of Counting in Vision-Language Models (arxiv.org)

Anytime-Valid Runtime Monitoring of Learned Barrier Certificates (arxiv.org)

CyberHull: Deterministic Malicious URL Isolation via Convex Hull (arxiv.org)

OptScientist: Automated Discovery of Optimization Laws (arxiv.org)

Agentic Ensemble with Vision-Language Models for Visual Anomaly Detection (arxiv.org)

Humans Correct Their Judgments Through Debate, but Weaker AI Models Do Not (arxiv.org)

Beyond 4 bits: Data-Centric Post-training Quantization and Recovery (arxiv.org)

A Radiation-Hardened, Fault-Aware Quantized Neural Network That Self-Heals (arxiv.org)

DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems (arxiv.org)

Sensitivity-Orthogonal Residual Learning for Adaptive Sideslip and Cornering Stiffness Estimation (arxiv.org)

Factual Access States in Diffusion Language Models: From Mechanism to Knowledge Editing and Retrieval (arxiv.org)

Navalia: Advancing Cowork Agents through Verifiable Task Synthesis and Long-Horizon Post-Training (arxiv.org)

RSI Control: Can AI Models Perform Covert Sabotage in the Self-Improvement Process? (arxiv.org)

How Language Models Organize Task Information (arxiv.org)

When Does Experience Memory Help Closed-Loop LLM Driving Agents? A Controlled Evaluation with Resampled Warmup Pools, Token-Matched Fillers, and Audited Baselines (arxiv.org)

How Fine-Tuning Objectives Shape Layer-Wise Information in LLM Hallucination Detection (arxiv.org)

E8-QAT: Efficient 2-Bit Quantization-Aware Training with Implicit Lattice Codebooks (arxiv.org)

EASformer: Environment-Action-State Alignment for Industrial Generative Ranking (arxiv.org)

Temporal Evidence Allocation For Longitudinal Clinical Retrieval: A Position And Technical Report (arxiv.org)

Reusing Multimodal Agents: Efficient Evaluation of Sound Designs in Games (arxiv.org)

Poisoning-Based Backdoors Are Not Cryptographic (arxiv.org)

Soliton Profiles in 1D: Enhanced Neural-Network Solvers and Constrained Deep Ritz Methods (arxiv.org)

Beyond the Prompt: Recovering Hidden Concepts in Text-to-Image Models (arxiv.org)

LoRASpace: A Pool-Wide Shared Substrate for Static and Dynamic Multi-LoRA Composition (arxiv.org)

Multi-Objective Evolutionary Semi-Supervised Learning for Resource-Aware Smart City Classification (arxiv.org)

Inferring continuous cognitive traits from limited neuroimaging data using pairwise convolutional neural networks (arxiv.org)

When Can LLM Reasoning Reliably Reuse Intermediate Results? (arxiv.org)

Does Multi-Agent Collaboration Pay When Agents Serve Users? Testing Agent Scaling in Dual-Control Settings (arxiv.org)

CUA-Recovery: Benchmarking Error Recovery in Long-Horizon Computer-Use Agents (arxiv.org)

SCODEL: Shared Canonical Ontology Definitions for Food Entity Linking and Ontology-Grounded Recipe Knowledge Graphs (arxiv.org)

Baseline and Incremental: User, Item, and Request Debiasing in Instagram Feed Ranking (arxiv.org)

The Router Already Knows (arxiv.org)

Learning the Geometry of Materials Space with Trained Neural Tangent Kernels (arxiv.org)

Steering intonational tunes in text-to-speech synthesis (arxiv.org)

Training Controllable World Models via Action Influence Maximization (arxiv.org)

Staged Continual Adaptation of Multimodal Foundation Models for Japanese Financial Documents (arxiv.org)

Observation Assimilation Transformer: Probabilistic Weather Analysis from Sparse Station Observations (arxiv.org)

Manifold-Informed Neural Networks for Low-Label ECG Electrolyte Inference (arxiv.org)

Recommending Bacterial Cultivation Conditions from 16S rRNA (arxiv.org)

Learning Normal Coordinates on Riemannian Manifolds with Flow Charts (arxiv.org)

Physical AI: Concepts, Techniques, and Challenges (arxiv.org)

M-Plan: Comprehensive Planning For LLM Agents with Memory Arrangement (arxiv.org)

Teaching AI Agent to Play Poker Strategically Through Preference Optimization (arxiv.org)

PW-Pruning: Phase-Aware Expert Selection for Speech-Centric Omni-Modal Models (arxiv.org)

TPO: Tri-level Distributionally Robust Learning for OOD Direct Preference Optimization (arxiv.org)

JAXbounce: A Continuous-Control Pong-Like Benchmark for Active Inference Beyond Low Dimensional Environments (arxiv.org)