AI News

LTM: Latent Transition Based In-Context Memory for Vision-Language-Action Models (arxiv.org)

ESG-Kor: A Korean Dataset for ESG-related Information Extraction and Practical Use Cases (arxiv.org)

PriorityFEC: Deadline-Aware Streaming FEC for KV-Cache Handoff in Edge Multi-Agent Systems (arxiv.org)

A Jury, Not an Electorate: Label-Free Reliability Estimation for Ensembles of Language Models (arxiv.org)

MIDI-REPA: Improving Text-to-Music Generation with Symbolic Representation Alignment (arxiv.org)

Coordination-Free Workload Regularization for Variable-Cost Distributed Training (arxiv.org)

Clean Data Can Still Carry Backdoors: Support-Persistent Backdoors for Model Reuse (arxiv.org)

What Should We Remove? Evaluating the Reliability and Costs of RLVR Data Decontamination (arxiv.org)

System1-Arena: Which Decisions Can Be Left to System 1? (arxiv.org)

PanoDiff-SR: Synthesizing Dental Panoramic Radiographs using Diffusion and Super-resolution (arxiv.org)

Matoi: Selective Revalidation of Consequence Evidence for Agent Authorization (arxiv.org)

When do quantum views help? Complementary classical-quantum representations in multi-view fusion (arxiv.org)

BiShield-TEE: On the (In-)Security of Unilateral Weight Obfuscation in On-Device TEE-Shielded LLM Partition (arxiv.org)

Independent Data-Integrity Verification for LLM-Driven Functional Coverage Closure in Hardware Design (arxiv.org)

Rethinking the Role of Tool Graph Memory in Multi-Turn Tool Use (arxiv.org)

Semantic Latent Alignment for Flow Matching with Invertible Bi-Lipschitz Adapters (arxiv.org)

When Should a Low-Light Pipeline Generate? Auditable Risk Control for Selective Generative Enhancement (arxiv.org)

L2GSSL: Label-Efficient Local-to-Global Self-Supervised Learning for Volumetric Healthcare Data (arxiv.org)

Calibration Language Shapes Block-Influence Pruning in Multilingual Language Models (arxiv.org)

AVALON: Zero-Shot Indoor Navigation with Human-Seeded Topological Memory (arxiv.org)

Decoding What iOS Location Services Send to Apple: An LLM Is an Unreliable Analyst but a Useful Toolmaker (arxiv.org)

Cross-Embodiment Causal Transfer from Human Videos to Robot Manipulation (arxiv.org)

Anvil: Scaling Embodied Spatial Intelligence through Knowledge Recirculation (arxiv.org)

Beyond Invariance: Factor-wise Selective Treatment of Shifts for Multimodal Learning (arxiv.org)

DuET: Note-Frame Dual Encoding for Piano Performance-to-Score Transcription (arxiv.org)

When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill (arxiv.org)

Rethinking Classifier-Free Guidance for TI2I Flow Matching (arxiv.org)

Reducing the Cost of Qudit State Preparation by Learning a Sparse Structure for SNAP Gates (arxiv.org)

Personalized Projected Ascent for Federated Stackelberg Optimization (arxiv.org)

Learning Rewards in Strategic Games (arxiv.org)

CodeHealth Predicts the Token Cost of Agentic Code Review: A Pre-Registered Study of 465 Agent-Authored Pull Requests (arxiv.org)

BridgeAfford: Bridging Query-Target Gap for Generalizable 3D Affordance Grounding (arxiv.org)

Agentic-Priors: Using Large Language Models to Address the Cold Start Problem in Preference Elicitation (arxiv.org)

Following Delegation Traces to Study Human Oversight in Interaction with Multimodal AI (arxiv.org)

PocketTSLM: Deploying an On-Device Soft-Prompt Time-Series Language Model for Privacy-Preserving Smartphone Reasoning (arxiv.org)

Same Spectrum, Harder First-Order Training: Phase-Locking in Single- versus Multi-Qubit Re-Uploading Circuits (arxiv.org)

Single-stage joint diffusion of sky images and clear-sky index for ultra-short-term irradiance forecasting: what the frame-to-irradiance stage discards (arxiv.org)

Priming Base Models for Agentic Post-Training: An Open Midtraining Recipe (arxiv.org)

DemoStream: Batched Task-and-Motion-Plan Optimization for Automated Learning from Demonstration (arxiv.org)

Can Capability Enhancement be Localized? (arxiv.org)

Spectrum Is Not Enough: A Random-Matrix Account of Data-Free Preconditioning for Differentially Private Learning (arxiv.org)

FlowBudget: Training-Free Sampling-Budget Estimation for Flow Matching Models (arxiv.org)

Expanding Attention: A Taylor-Series Route to Linear-Time Transformers with a Fixed Number of Recurrent States (arxiv.org)

Sensor Diagnosis for Fuel and Emissions Monitoring in Heavy-Duty Diesel Vehicles (arxiv.org)

Resisting Harmful Instructions through Training-Free Weight Editing (arxiv.org)

Lean Graph Search for Formal Theorem Proving (arxiv.org)

Rethinking Routing Exploration in Modern Mixture-of-Experts (arxiv.org)

Learning to Read Out: Unembedding Dynamics in Language Model Pretraining (arxiv.org)

Behavioral Compositionality of Sparse Autoencoder Representations (arxiv.org)

Rethinking Learning from Label Proportions via Moment Matching (arxiv.org)