AI News

ReText: Text Boosts Generalization in Image-Based Person Re-identification (arxiv.org)

Self-Refining Vision Language Model for Robotic Failure Detection and Reasoning (arxiv.org)

Open-world AI: a Tale of Three Requirements (arxiv.org)

ARIA: Runtime Inspection and Amendment for Multi-Agent Systems (arxiv.org)

When Reality Outpaces Cognition: A New Scaling Principle for Reliable Intelligence (arxiv.org)

L2G-SR: Learning to Guide LLM-based Symbolic Regression (arxiv.org)

IneqAgent: Agentic Inequality Proving in Lean with Symbolic Computation (arxiv.org)

Memory as Weights: Internalizing Long-Term History for Streaming Videos (arxiv.org)

Multi-SFT: Permissively-Licensed Multilingual Instruction-Following (arxiv.org)

3DCodeVerse: Exploring 3D Worlds as Code (arxiv.org)

SIEVE: Overcoming Topological Obstruction in Equivariant Self-Supervised Learning (arxiv.org)

Bellman-Consistent Trajectory RL for LLM Arithmetic Reasoning (arxiv.org)

SALTO: Solver-Anchored Lifted Transfer Operators for Nonlinear Event Inference (arxiv.org)

Vertical Sliced Wasserstein Distances for Meta-Measures (arxiv.org)

Witness Overlap: Directional Provenance Inside Open-Weight Model Families (arxiv.org)

Quantile-Guided Diffusion for Probabilistic Time-Series Forecasting (arxiv.org)

Counterfactual Neighborhood Risk Fields: Evidence-Grounded Selective Safety for Vision-Language Models (arxiv.org)

When Correctness Feedback Erases Evidence: Phase-Separated Completeness Auditing in Verus (arxiv.org)

SkillNEO: Non-Elitist Evolutionary Optimization of Agent Skills (arxiv.org)

The Delegation Dilemma: Economies of Scale, Accountability, and Polycentric Governance (arxiv.org)

Isolating a Language Model's Judgment with a Label-Free Subspace (arxiv.org)

MARS: A Momentum-Aware Adaptive Rating System for Personalized Assessment (arxiv.org)

When Does "Post-Call" Begin? An Availability Contract for Earnings-Call NLP Benchmarks (arxiv.org)

Reliable Knowledge Acquisition for Continual Learning Agents (arxiv.org)

Retrieved but Not Read: Understanding Why Search-Augmented LLMs Fail to Use Available Evidence (arxiv.org)

FIRM: Fisher-Informed Retention Merging for Federated LLM Unlearning (arxiv.org)

Encoded but Exposed: Language Embeddings can Reveal More Than They Should in Recommender Systems (arxiv.org)

Human Interventions as Labels: Localizing, Steering, and (Sometimes) Editing the Computations Behind LLM Agent Failures (arxiv.org)

Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning (arxiv.org)

Qwen-Audio-Jev: A Unified Interface for Audio Decisions (arxiv.org)

Can LLMs Tell Stories Out of Order? A Corpus and Benchmark for Nonlinear Narrative Generation (arxiv.org)

Trajectory-Level Speculation with Semantic Verification for Tool-Using Agents (arxiv.org)

Can Visual Generation Expand What Unified Multimodal Models Know? (arxiv.org)

GeoState: Learning View-Decoupled Geometry States for 3D World Modeling (arxiv.org)

ContextLift: Breaking the Spurious Correlation Behind Conditional Emergent Misalignment (arxiv.org)

DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents (arxiv.org)

HyCo: Hybrid Drafting for Cost-Efficient Mobile-Cloud Speculative Decoding (arxiv.org)

Policy-Aligned Predictive Verification for Adaptive VLA Action-Chunk Execution (arxiv.org)

ManiFusion: Unlocking the High-Throughput Generation via Superposition in Manifold Space (arxiv.org)

Full-Duplex Interaction with Speech, Expressions and Gestures (arxiv.org)

X-Fly: Learning to Reach and Stop in Full-Chunk Aerial Vision-Language Navigation (arxiv.org)

Where, How, and When: Query-Guided Grounding for Video Understanding (arxiv.org)

RecSoup: Efficient Model Merging Approach for Controllable Multi-Objective Recommendation (arxiv.org)

FedSumUp: User-Auditable Privacy-Aware Representation Sharing for Resource-Constrained Federated Intelligence (arxiv.org)

Judging the Judges: Confidence and Disagreement in Rubrics as Rewards (arxiv.org)

Language Models Share a Readout Bottleneck in Working Memory (arxiv.org)

CARE-SD: Explainable Reasoning Architecture for Local SLM Self-Disclosure Analysis in Textual Data (arxiv.org)

Pathwise Shrinkage and Probabilistic Calibration of the Multiplicative Half-Cauchy Process (arxiv.org)

Adaptive Randomized Neural Networks for PDEs: Structure-Aware Feature Redistribution and Residual-Driven Block Enrichment (arxiv.org)

SHIP: Structured and Hierarchical Insight with Privacy (arxiv.org)