AI News

TiWeaver: Unified Temporal Dynamics Modeling via Contextual Patching (arxiv.org)

When to Re-Plan: Subgoal Persistence in Hierarchical Latent Reasoning (arxiv.org)

Attribution via Distributional Paths for Information Revelation (arxiv.org)

Diagnosing Knowledge Gaps in LLM Tool Use: An Agentic Benchmark for Novel API Acquisition (arxiv.org)

SCOPE: Real-Time Natural Language Camera Agent at the Edge (arxiv.org)

Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning (arxiv.org)

Towards Non-Monotonic Entailment in Propositional Defeasible Standpoint Logic (arxiv.org)

AugMask: Training Diffusion Models on Incomplete Tabular Data via Stochastic Augmentation and Masking (arxiv.org)

Enhancing Protein-Protein Interaction Prediction with Hierarchical Motif-based Multimodal Protein Embedding (arxiv.org)

WRIT: Write-Read Intensive Trajectory Synthesis for Multi-Turn User-Facing Agents (arxiv.org)

IdEst: Assessing Self-Supervised Learning Representations via Intrinsic Dimension (arxiv.org)

FederatedSkill: Federated Learning for Agentic Skill Evolution (arxiv.org)

Resource-Constrained Adaptive Inference for Sequential Pricing (arxiv.org)

Too Much of a Good Thing: When sim2real Efforts Impede Policy Learning (And What to Do About It) (arxiv.org)

Constitutional On-Policy Safe Distillation (arxiv.org)

PURGE: Projected Unlearning via Retain-Guided Erasure (arxiv.org)

State-Coupled Volatility in Latent Dynamical Systems: Recovery Under Partial Observation (arxiv.org)

SEAOTTER: Sensor Embedded Autoencoding with One-Time Transcode for Efficient Reconstruction (arxiv.org)

Glass Box at Orbit: A Constitutional AI Verification Framework for Trustworthy Autonomous CubeSat Intelligence (arxiv.org)

OpenAgenet/OAN: Technical Architecture for Trust-Governed Agent Identity and Discovery (arxiv.org)

EvoDrive: Pareto Evolution for Safety-Critical Autonomous Driving via Self-Improving LLM Agents (arxiv.org)

DiffUNet^2: Bidirectional Prediction, Probabilistic Generation and Collaborative Visual Discovery for Scientific Data (arxiv.org)

Echelon: Auditable Aggregate-Only Language-Model Adaptation Across Privacy Boundaries (arxiv.org)

When RLHF Fails: A Mechanistic Taxonomy of Reward Hacking, Collapse, and Evaluator Gaming (arxiv.org)

Synthetic Hallucinations, Real Gains: Hard Negatives from Frontier Models for FIM Hallucination Mitigation (arxiv.org)

TIDFormer: Exploiting Temporal and Interactive Dynamics Makes A Great Dynamic Graph Transformer (arxiv.org)

From 'What' to 'How' and 'Why': Sharing LLM-Generated Retrospective Summaries of Older Adults' Passive Tracking Data with Remote Family Members (arxiv.org)

Consistency Training Can Entrench Misalignment (arxiv.org)

Taiji: Pareto Optimal Policy Optimization with Semantics-IDs Trade-off for Industrial LLM-Enhanced Recommendation (arxiv.org)

Coherent Swap Regret and Channel-Proof Learning (arxiv.org)

VistaHop: Benchmarking Multi-hop Visual Reasoning for Visual DeepSearch (arxiv.org)

OpenAgenet/OAN: Open Infrastructure for Trusted Agent Interconnection (arxiv.org)

Conditional Latent Diffusion Model with Fourier-based Motion Modelling for Virtual Population Synthesis (arxiv.org)

Fairness Definitions and Metrics in Deep Reinforcement Learning for Drug Discovery in Healthcare: A Rapid Evidence Review (arxiv.org)

GuidedBridge: Training-freely Improving Bridge Models with Prior Guidance (arxiv.org)

ROBUST-WT: Robust Uncertainty-aware Segmentation Transform via Whitening and Training Enhancements (arxiv.org)

Reinforcement Learning from Cross-domain Videos with Video Prediction Model (arxiv.org)

One Transit Is All You Need: Detecting Exoplanets Through Learned Stellar Behaviour with EXOVEIL (arxiv.org)

$\mathbb{R}^{2k}$ is Theoretically Large Enough for Embedding-based Top-$k$ Retrieval (arxiv.org)

Can Local Learning Match Self-Supervised Backpropagation? (arxiv.org)

Finding Kissing Numbers with Game-theoretic Reinforcement Learning (arxiv.org)

Fast and Expressive Multi-Byte Prediction with Probabilistic Circuits (arxiv.org)

Lethe: Adapter-Augmented Dual-Stream Update for Persistent Knowledge Erasure in Federated Unlearning (arxiv.org)

TriEval: A Resource-Efficient Pipeline for LLM Bias, Toxicity, and Truthfulness Assessment (arxiv.org)

When Helping Hurts and How to Fix It: Multi-Agent Debate for Data Cleaning (arxiv.org)

Assistax: A Multi-Agent Hardware-Accelerated Reinforcement Learning Benchmark for Assistive Robotics (arxiv.org)

Quantifying Faithful Confidence Expression in Large Reasoning Models (arxiv.org)

Planning with Uncertainty: Symmetries, Policy Inference, and Solution Compression (arxiv.org)

AlignAtt4LLM: Fast AlignAtt for Decoder-Only LLMs at IWSLT 2026 Simultaneous Speech Translation Task (arxiv.org)

Agentic Chain-of-Thought Steering for Efficient and Controllable LLM Reasoning (arxiv.org)