AI News

AdaC-GRPO: Towards Mastering Long-Horizon Visual Sudoku Puzzles via Adaptive Curriculum Learning (arxiv.org)

Asymptotic Analysis of Discrete-Time Self-Attention Dynamics in Transformers (arxiv.org)

Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning (arxiv.org)

Target-Constrained Pooling for Trajectory Extrapolation (arxiv.org)

Towards Easier-to-Tune Training Algorithms: New Results from AlgoPerf (arxiv.org)

CB-Orchestrator: Adaptive Workflow Optimization for LLM Agents via Contextual Bandits (arxiv.org)

Specialists Learn Faster: Knowledge Organization in Pre-training Shapes Knowledge Update Efficiency (arxiv.org)

JANUS: Enhancing Long-Horizon Autonomous Research with Co-Evolving World Models (arxiv.org)

Forgetting is Not Always Bad: A Neuro-Inspired Memory Repair Mechanism for Poisoned LLM Agents (arxiv.org)

BoundaryEvo: Data-Model Co-Evolution at the Capability Boundary for Long-Chain Reasoning Distillation (arxiv.org)

Masked Diffusion Language Agents for Tool-Integrated Chemical Reasoning (arxiv.org)

AlphaNIST: Emergent Assembly Skills from Binary Task Rewards (arxiv.org)

Returning to Development States: When Full Sandbox Snapshots Help Coding Agents (arxiv.org)

$L_0$ Feature Selection via Monte Carlo Tree Search in High-Dimensional Financial Regression (arxiv.org)

EchoTab: Learning Measurement-Grounded Representations for Echocardiographic Diagnosis through Tabular Reconstruction (arxiv.org)

Private and Distributionally Robust Federated Multiclass Classification (arxiv.org)

PANDA: Prior-guided Attentional Dual-path Architecture (arxiv.org)

TabReliBench: Evaluating Tabular Foundation Models Beyond Predictive Accuracy (arxiv.org)

Evolving Creatives with Clicks: CTR-Driven Advertising image Editing via Attribute Disentanglement and Preference Alignment (arxiv.org)

Beyond Actions: Learning Future Operation Targets for Vision-Language-Action Models (arxiv.org)

How Reliable are AMR Parser Outputs? Uncertainty Estimation for AMR Parsing (arxiv.org)

Agent4Spec: An LLM-based Agentic System to Optimize API Specifications through Automated Validation and Market Analysis (arxiv.org)

PRIME: Pre-Retrieval Index Modeling Enhancement for Structured Document Retrieval-Real (arxiv.org)

REDaCT: Collapse-Free Unlearning in Large Reasoning Models (arxiv.org)

On the Encoding of Calendar Features in Covariate-Informed Time Series Foundation Models (arxiv.org)

VQIL: Visible Quantum Interference Learning for Perturbation Response Ranking on Graphs (arxiv.org)

Transformer-based In-Context Score Operator Learning for Diffusion Models (arxiv.org)

Variance-Reduced Policy Gradient is Nearly Minimax-Optimal (arxiv.org)

Inject a Turn, Not a Token: Learning Where to Redirect LLM Reasoning for Self-Correction (arxiv.org)

Agreement Without Alignment: Label-Free Rubric Clarification Increases LLM-Judge Consensus but Not Agreement with Humans (arxiv.org)

AgentSWE: Can Coding Agents Build the Agent You Actually Want? (arxiv.org)

ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents (arxiv.org)

CNN-Transformer with Sigmoid Label for High-Resolution Sub-Degree DOA Estimation (arxiv.org)

Meta-Reflective Evolution Effectively Improves Production AI Systems (arxiv.org)

Towards Understanding the Training of GFlowNets: A Theoretical Perspective (arxiv.org)

Machine Unlearning for Federated Socially Fair k-Means Clustering (arxiv.org)

Amortized Counterfactual Prediction with Graph Neural Networks (arxiv.org)

SeoulMMOD: A Large-Scale Multimodal Origin-Destination Flow Benchmark (arxiv.org)

Semantic Views: Certified Reuse of LLM Labels Across Semantic Queries (arxiv.org)

Orthros: Phase-Aware Heterogeneous Attention for Efficient Transformers (arxiv.org)

SafeDiSTL: Decomposing Joint Constraints for Safe Multi-Agent Reinforcement Learning (arxiv.org)

Studying the Effects of Task Diversity Using Small Mathematical Worlds (arxiv.org)

Equal-Budget Monte Carlo Tree Search versus Randomized-Restart List Scheduling for DAG Workflows: A Controlled Equal-Budget Computational Study (arxiv.org)

Can Language Models Play the Wikipedia Game? (arxiv.org)

RATED: Reliability-Aware AI Agent Tool Retrieval via Execution Dynamics (arxiv.org)

X-AgentRed: Red-Teaming Personal Agents in Evolving Multi-Service Ecosystems (arxiv.org)

Studying Comics at Scale with Artificial Intelligence: The PAGE Framework and Three Studies of American Comics and Japanese Manga (arxiv.org)

Primal-Dual Flow Matching for Sample-Wise Constrained Generation (arxiv.org)

SynAct: Synergistic Active Feedback Learning for LLM Agents (arxiv.org)

Machine Learning and Time-Series Foundation Models for Day-Ahead Electricity Forecasting in a University Building (arxiv.org)