AI News

Target Updates May Stabilize Linear Q-Learning: Periodic and Soft Dynamics (arxiv.org)

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

Trans GAN-WT: A Feature Extraction and Interactive Learning-Based Anomaly Detection Model for Wind Turbine Time Series Data (arxiv.org)

A Training-Free Mixture-of-Agents Framework for Multi-Document Summarization using LLMs and Knowledge Graphs (arxiv.org)

Strongly Polynomial Time Complexity of Policy Iteration for $L_\infty$ Robust MDPs (arxiv.org)

Position: Prioritize Identifying Structure, Not Complex Models, for Scientific Discovery (arxiv.org)

Calibrating Urban Traffic Simulation from Sparse Road Observations via Genetic Optimization (arxiv.org)

BigFinanceBench: A Workflow-Grounded Benchmark for Financial-Research Agents (arxiv.org)

Re-Evaluating Continual Learning with Few-Shot Adaptation (arxiv.org)

EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management (arxiv.org)

Leveraging BART to Assess CS1 C++ Programming Assignments using Rubric-based Criteria (arxiv.org)

P\textsuperscript{2}-DPO: Grounding Hallucination in Perceptual Processing via Calibration Direct Preference Optimization (arxiv.org)

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

SketchSong: Hierarchical Song Generation with Sketch Planning and Fine-Grained Multi-Track Modeling (arxiv.org)

Self-Soupervision: Cooking Model Soups without Labels (arxiv.org)

Enhancing Operational Safety via Agentic Dialogue Hazard Identification Analysis (arxiv.org)

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

From Control Boundary to Insurance Claim: Reconstructing AI-Mediated Losses Through the CER Framework (arxiv.org)

Pretraining Language Models on Historical Text (arxiv.org)

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

Dynamic Short Convolutions Improve Transformers (arxiv.org)

Fast Organic Crystal Structure Prediction with Unit Cell Flow Matching (arxiv.org)

LLM-Assisted Reranking to Operationalize Nuanced Objectives in Recommender Systems (arxiv.org)

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

DECA: Decentralizing Block-Wise Adam for Efficient LLM Full-Parameter Fine-Tuning on Non-IID Data (arxiv.org)

Bayesian Tensor Decomposition with Diffusion Model Prior (arxiv.org)

Multi-Modal Machine Learning for Breast Cancer Recurrence Prediction (arxiv.org)

Learning Temporal Causal Structure via Smooth Differentiable Optimization (arxiv.org)

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

The DeepSpeak-Agentic Dataset (arxiv.org)

Hand Trajectory Fusion for Egocentric Natural Language Query Grounding (arxiv.org)

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

Neural Attention Search Linear: Towards Adaptive Token-Level Hybrid Attention Models (arxiv.org)

Dynamic Objective Selection with Safeguards and LLM Oversight for Financial Decision-Making (arxiv.org)

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

A Cartesian-3j Framework for Machine Learning Interatomic Potentials (arxiv.org)

Towards Blind Lens Aberration Correction via Large LensLib Pre-training and Discrete Degradation Priors (arxiv.org)

Learning to See via Epiretinal Implant Stimulation in silico with Model-Based Deep Reinforcement Learning (arxiv.org)

Rethinking Neural Width for Alternating Current Optimal Power Flow Proxies (arxiv.org)

FGRPO: Federated GRPO with Adaptive Aggregation on Non-IID Data (arxiv.org)

ScoreStop: Gradient-based early stopping using functional score tests (arxiv.org)

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

FORGE: Multi-Agent Graduated Exploitation and Detection Engineering (arxiv.org)

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

GFFMERGE: Efficient Merging of Graph Neural Force Fields and Beyond (arxiv.org)

PerchRL: Vision-Based Agile Perching on Inclined Platforms under Rapid and Irregular Motion (arxiv.org)

Low-Frequency Shortcuts in Texture-Driven Visual Learning (arxiv.org)

q0: Primitives for Hyper-Epoch Pretraining (arxiv.org)

dstack-capsule: Pod-Level Remote Attestation for Confidential Workloads on Kubernetes (arxiv.org)

Constitutional On-Policy Safe Distillation (arxiv.org)