AI News

Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents (arxiv.org)

Consensus Group Relative Policy Optimization for Text Generation (arxiv.org)

ReCAST: Restoration-aware Cascaded Stage-wise Training for Obfuscated SMS Risk Classification (arxiv.org)

SimFuse3D: Source-Guided Target Simulation and Confidence-Guided Multi-Stage Localization Reweighting for Cross-Platform 3D Object Detection (arxiv.org)

Attention-guided super-resolution of 4D flow MRI in carotid arteries (arxiv.org)

RefactorPlatform: An Open-Source Harness for Controlled Evaluation of Repository-Scale Refactoring Agents (arxiv.org)

Train What You Deploy:Token-Faithful Post-Training of a Production Coding (arxiv.org)

Multi-Modal Time Series Prediction via Mixture of Modulated Experts (arxiv.org)

GLOW: Graph-Language Co-Encoding for Agentic Workflow Performance Prediction (arxiv.org)

Forecast Skill Is Not Decision Skill: Evidence from Weather-Dependent Decision Tasks (arxiv.org)

Fractal and Chaotic Activation Functions in Echo State Networks: Preprocessing Topology Governs the Echo State Property (arxiv.org)

How to Speculate about Uncertainty in Agentic Coding? A Draft-Model Gate Method (arxiv.org)

Partial Inverse Design of High-Performance Concrete Using Cooperative Neural Networks for Constraint-Aware Mix Generation (arxiv.org)

ResLearn-XR: Residual Learning for Network Traffic and Quality-of-Experience-Aware Modeling in Extended Reality (arxiv.org)

AlcaTRAz - Anchored Tree-Rule Defense Against Jailbreaks (arxiv.org)

A Deep Generative Model for Synthesizing Labeled Wireless Signals (arxiv.org)

Molecular D\'ej\`a Vu: Digit-Level Retrieval of Published Values in Frontier Language Models (arxiv.org)

CUA-Universe: A Scalable and Dynamic Environment for Hybrid GUI+CLI Agents (arxiv.org)

Who Should Grade My Work? Student Perspectives on Transparent AI-Assisted Writing Assessment in Higher Education (arxiv.org)

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability (arxiv.org)

Technical Manual for a Toolkit for Measuring Contextual Individuation in Transformer Language Models (arxiv.org)

LLM-Driven Algorithm Design for Quantum Circuit Synthesis based on Binary Decision Diagrams (arxiv.org)

Large Language Models for HVAC Operations in Building Energy Systems: A Critical Review of Methods, Applications, and Deployment Readiness (arxiv.org)

RISE: Recursive Improvement via Self-Extrapolating Policy Distillation (arxiv.org)

Substrate-Aware AI Agents: Execution Context as a First-Class Input (arxiv.org)

ACE: Adaptive Calibration-Free Expert Skipping for MoE-based LLMs (arxiv.org)

Necessary or Sufficient? Evaluating LLM Explanations With Behavioural Evidence (arxiv.org)

CABAL: Multi-Agent Simulacra for Tracing the Effects of Collusive Bidding in Peer Review (arxiv.org)

What Matters in On-Policy Distillation? A Perspective on Data Efficiency and Data Selection (arxiv.org)

The Mirror Agent Model: a Bayesian Architecture for Interpretable Agent Behavior (arxiv.org)

SciDocBench: A Workflow-Centered Benchmark and Data Pipeline for Scientific Document Understanding (arxiv.org)

Unifying ICL, SFT, KL-Regularized RL Through a Bayesian Lens (arxiv.org)

Compact Bellman-Grounded Cognitive Maps for Cost-Aware Navigation (arxiv.org)

ProCA: Progressive Contrastive Alignment for Robust EEG Visual Decoding (arxiv.org)

MM-IFEval-Pro: A Multilingual and Attack-Resistant Benchmark for Instruction-Following in Vision-Language Models (arxiv.org)

ElderBench: Benchmarking Autonomous Mobile Agents for Older Adults (arxiv.org)

Long Horizon Transformer Quantile Fault Prediction for Multi Site Industrial Predictive Maintenance (arxiv.org)

CPR-IE:A Compression-Prediction-Resource Intelligence Efficiency Metric (arxiv.org)

When Financial Fine-tuning Fails: A Three-Level Detectability Analysis of Numerical Hallucination in Domain-Adapted Language Models (arxiv.org)

MedFlow: Class-Aware Multi-Scale Generation for Medical Time-Series Synthesis (arxiv.org)

Hierarchical Possession-Aware Graph Pointer Network for Pass Receiver Selection (arxiv.org)

Whose record is this? Diagnosing and authorizing record use in personalized multimodal models (arxiv.org)

ProtLingo: Efficient Protein Language Modeling via Conditional Memory and Expert Routing (arxiv.org)

Continual Graph Memory for Adaptive Recommendation under Intent Drift (arxiv.org)

A Cost-Aware Agentic Architecture for NL-to-SQL over Nested Enterprise Schemas, with a New Benchmark (arxiv.org)

Leveraging Imperfect Restoration for Data Availability Attack (arxiv.org)

$\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction (arxiv.org)

Does the Selected Object Reach the Reader? Auditing Identity Handoffs in Grounded Language-Model Pipelines (arxiv.org)

Extremely Sparse Supervision Incentivizes Reasoning Ability (arxiv.org)

La Agente \'Optima: Towards Agentic Self-Driving Laboratories (arxiv.org)