AI News

Subgraph-Aware Training of Language Models for Knowledge Graph Completion Using Structure-Aware Contrastive Learning (arxiv.org)

When Messages Go Wrong: Understanding Communication Faults in LLM-Based Multi-Agent Systems (arxiv.org)

Security in LLM-Assisted Coding: A Systematic Literature Review (arxiv.org)

The Imperishable Record, Forgiveness, and the Future of Criminal Justice (arxiv.org)

CacheLink: Functional KV Transport for Zero-Prefill Cross-Model Context Handoff (arxiv.org)

RealityBench: Simulated Environments to Evaluate a Vision Language Model's Ability to Document Human Physical Work (arxiv.org)

REPRO: Reflective Evolutionary Prompt Optimizer for Automated LLM Red-Teaming (arxiv.org)

SARATHI: Decoupled Data-Free Probing and Adaptive Reconstruction for Structured LLM Pruning (arxiv.org)

OceanFlow: Coherent Flow Matching for Probabilistic Ocean Modeling (arxiv.org)

Test-Time Scaling via Budgeted Multi-Attribute Verification (arxiv.org)

Reinforcement Learning In Partially Verifiable Environments for Colon Cancer Treatment Planning (arxiv.org)

World-T3: Test-Time Training as Persistent Memory for Real-Time Video World Models (arxiv.org)

Know-Thy-Not: Training-Free Targeted Negation in Dual-Encoder Vision-Language Models (arxiv.org)

Foundation Models for Event Prediction Across Heterogeneous Event Logs (arxiv.org)

The belief-action gap in language model agents (arxiv.org)

Compositional RL: Evaluating and Training LLMs for Collaborative Reasoning (arxiv.org)

Assessing the Performance of AI Models in Misinformation Detection Using Predefined Social Noise Constructs (arxiv.org)

VeriPIE: Checkable Query Plans and Evidence-Guided Selection for Text-to-SQL (arxiv.org)

YUBI-Real2Sim: 3DGS Scene Verification for Reliable Policy Evaluation (arxiv.org)

Enhanced Detective: A Dual-Stage Proactive Defense Against Internal Poisoning Attacks in Financial Fraud Detection Using Federated Learning (arxiv.org)

Box-WAM: Box Trajectories as Compact State Representations for World Action Models (arxiv.org)

Towards Green AI: Evaluating Energy Efficiency for Quantized LLM Inference (arxiv.org)

Convex-Analytic View of Query Selection for Safety Verification of Markov Decision Processes (arxiv.org)

Many Steps into One: Reasoning Contraction and the Limits of Safe Memory Reuse for LLM Agents (arxiv.org)

FlashVSR+: Efficient Video Super-Resolution via Hybrid Attention and Reward-Yield Distribution Matching (arxiv.org)

Benchmarking Self-Supervised 3D Medical Vision Foundation Models and Transferability Estimation (arxiv.org)

Dissociating Answerability from Extraction in LLM Question Answering over Radiology Reports (arxiv.org)

Chasing Label Shifters: A Change-Aware Framework for Dynamic Graph Node Classification (arxiv.org)

HelpMeOut: A Neuropsychology-Grounded Benchmark for Error-Aware Monitoring of Household Task Execution (arxiv.org)

Taming Softmax at the Limits of Muon: A Unified Approach through Hilbert Geometry (arxiv.org)

Revisiting the 16x Pretraining Compute Gap between Diffusion and Autoregressive Language Models (arxiv.org)

Tanawpamana: Benchmarking Vision-Language Models on Philippine Cultural Heritage Sites (arxiv.org)

Morphology-aware deep learning predicts colorectal cancer survival from whole-slide images (arxiv.org)

4LC: 4D Latent Cache for Efficient Conditioning in Dynamic World Modeling (arxiv.org)

The Tokens Strike Back: Safety-Aware Denoising with Directional Embedding-based Revision (arxiv.org)

DeltaTTG: Training Multi-Turn Agents with a Token-to-Go Progress Meter (arxiv.org)

PrT-Flow: Probability-Tractable Flow Matching for Reinforcement Learning (arxiv.org)

Dynamic K-Head: Difficulty-Supervised Budget Selection for Continuous Latent Reasoning (arxiv.org)

LatentRiskBench: Measuring State-Dependent Risks in Enterprise Agents (arxiv.org)

zPocket: Better Pocket Finding with Frozen Co-Folding Features (arxiv.org)

LeanPRM: Lean-Grounded Process Supervision for Natural Language Proof Verification (arxiv.org)

From Token Correction to Claim Admission: Training-Free Hallucination Mitigation for Large Vision-Language Models (arxiv.org)

Where Small Mistakes Matter: Quantifying State Criticality for Imitation Learning (arxiv.org)

TSRL: Trigger-based Supervised Reinforcement Learning for Continual Agent Training (arxiv.org)

SkillScreen: High-Recall Screening of Malicious Agent Skills with Decision Models (arxiv.org)

CoCA: Criterion-Aware Credit Assignment for Rubric-Based Reinforcement Learning (arxiv.org)

Explore with Memory, Learn without It: In-Context Exploration and Self-Distillation for LLM Reasoning (arxiv.org)

Train for Many, Update with One: Improving Multi-Sample Reasoning via Fenchel-Bregman Posterior Learning (arxiv.org)

HiSTOR: Structured Treatment-outcome Representation Learning for High-Dimensional Longitudinal Data (arxiv.org)

Auditing Failure Attribution: Label Predictability and Judge Behavior (arxiv.org)