AI News

Regularized Small Area Estimation with Graph Laplacian Benchmarking priors (arxiv.org)

When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization (arxiv.org)

Tight Bounds for Equivalence Testing with Non-Adaptive Conditional Samples (arxiv.org)

Decision-Sufficient Posterior Approximation (arxiv.org)

Sequential Conditional Independence Testing with Machine Learning Models (arxiv.org)

SPD-MetaFormer is what you need for small-data brain decoding (arxiv.org)

Stochastic Teacher Intervention for Agentic On-Policy Distillation (arxiv.org)

Grammar Concept Annotation at Scale: Deployed Fine-Tuned Small Language Models Outperform Prompted Frontier Models (arxiv.org)

VICO: Visual Environments Co-Evolving for Vision-Language Model Reasoning (arxiv.org)

NOMOS: Compiling Written Policies into Statically Verified Tool-Call Gates for LLM Agents (arxiv.org)

A Graph Neural Network for Global Daily Fire Radiative Power Prediction at Medium-Range Lead Times (arxiv.org)

Mid-Training Language Models on Raw Video (arxiv.org)

SFT-as-Context Mitigates Forgetting in Supervised Fine-Tuning (arxiv.org)

BRANCH: Bypassing Multi-Scanner AI Guardrails (arxiv.org)

LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC (arxiv.org)

SpaceFlow: Locally Controllable 3D Generation (arxiv.org)

ARC: A Reasoning Recipe for Robot Foundation Models (arxiv.org)

Distilling Routed 3D Privilege for Spatial Reasoning in Vision-Language Models (arxiv.org)

MetaEncoder: Exploring the Limit of Bi-Encoders for Multimodal System One Decision Making with Natural Language Interface (arxiv.org)

Region-Aware CLS Token Augmentation for Fine-Grained Image Retrieval (arxiv.org)

Omni-Diffusion-Distill: Few-Step Distillation of Unified Multimodal Diffusion Large Language Models (arxiv.org)

AI4Fire: Evaluating Large Language Models on Wildfire Tasks (arxiv.org)

GHARP: Real-time Gaussian Head Animation from Large-scale Reconstruction Prior (arxiv.org)

Explaining the Saliency Map Sparsity of Adversarially-Trained Neural Networks (arxiv.org)

ContiLNN: Mitigating Slice Sampling Discontinuity with Liquid Neural Networks for Medical Image Restoration (arxiv.org)

Multi-Agent Egocentric World Model with Fine-Grained Embodied Interaction (arxiv.org)

Machine Learning Meets High-Energy Nuclear Physics: From Pattern Recognition to Physics-Integrated Discovery (arxiv.org)

DVD: Dynamic Vector Decoding for Efficient MLLM-based Perception (arxiv.org)

Reconstruction of Multiscale Plasma Dynamics Across Operating Regimes (arxiv.org)

Higher-Order Morphology Priors for Quadruped Reinforcement Learning Under Actuator Degradation (arxiv.org)

Implementation Guidelines for Data Quality Metrics (arxiv.org)

When Listening Becomes Easier: Scrubbing Visual Cues for Shortcut-Free VLAs (arxiv.org)

Power Side-Channel Membership Inference Attack on Embedded Machine Learning (arxiv.org)

Randomized Transport Maps for Model-Free Policy-Gradient Mean-Field Control (arxiv.org)

Conversational Voice Aesthetic Model with Reinforcement Learning from Human Listeners (arxiv.org)

A mesh-based neural energy method for the simulation of heterogeneous composites (arxiv.org)

Enabling Preference-driven Unlearning in Few-step Distilled Text-to-Image Diffusion Models (arxiv.org)

Syn-Omni: Structured Specialization and Progressive Collaboration for Omnimodal Embeddings (arxiv.org)

How Much Audio Is Left In An Embedding? An Inversion Audit Of Audio Encoders (arxiv.org)

Real-Time Motion Planning with Dynamic Hazards: Classical vs. Learning-Based Methods (arxiv.org)

La-Ribo: RNA Co-Design via Geometry-Latent Flow Matching (arxiv.org)

ReSI: Recursive Safety Improvement toward Resistant and Resilient AI (arxiv.org)

Characterizing Overconfident Failure in LLM-Based Code Generation (arxiv.org)

From Prompting to Composing: A Spatial Canvas Interface for Poster Generation (arxiv.org)

VibeEdit: Image Editing with Canvas Instructions (arxiv.org)

AI-Based On-Board Maritime Object Detection for Earth Observation Payload Data Reduction on Versal Embedded Hardware (arxiv.org)

Unifying Policy Learning and State Prediction through Spatial Language Modeling (arxiv.org)

RFChipAgent: Multi-Agentic AI Flow for Analog/RF Chip Design (arxiv.org)

Real Long-Term Memory for AI: A 50-Million-Token Window That Is Faster and Cheaper Than Recompute (arxiv.org)

Diagnosing and Recovering from Observation-Space Shift at Long-Horizon Skill Seams (arxiv.org)