AI News

FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems (arxiv.org)

VioLA: Learning Generalist Humanoid Control Policies from Human Data (arxiv.org)

One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts (arxiv.org)

A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control (arxiv.org)

CSF: Contextual Safety Filtering for Motion Generators (arxiv.org)

Class Machine Unlearning for Complex Data via Concepts Inference and Data Poisoning (arxiv.org)

C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models (arxiv.org)

BrainATCL: Adaptive Temporal Brain Connectivity Learning for Functional Link Prediction and Age Estimation (arxiv.org)

MedMemoryBench: Benchmarking Agent Memory in Personalized Healthcare (arxiv.org)

GLM-RAG: Graph Language Models for Graph-Based Retrieval-Augmented Generation (arxiv.org)

GPF-Net: Gated Progressive Fusion Learning for Polyp Re-Identification (arxiv.org)

SRUG: A Fusion-Driven Generator Network for Medical Image Translation (arxiv.org)

Rethinking Latency Denial-of-Service: Attacking the LLM Serving Framework, Not the Model (arxiv.org)

Fractional Heat Kernel for Semi-Supervised Graph Learning with Small Training Sample Size (arxiv.org)

Local Timescale Gates for Timescale-Robust Continual Spiking Neural Networks (arxiv.org)

Explanation Multiplicity in SHAP: Characterization and Assessment (arxiv.org)

Trust, Don't Trust, or Flip: Robust Preference-Based Reinforcement Learning with Multi-Expert Feedback (arxiv.org)

\$OneMillion-Bench: How Far are Language Agents from Human Experts? (arxiv.org)

Explanation Multiplicity in SHAP: Characterization and Assessment (arxiv.org)

Obscuring Data Contamination Through Translation: Evidence from Arabic Corpora (arxiv.org)

Trust, Don't Trust, or Flip: Robust Preference-Based Reinforcement Learning with Multi-Expert Feedback (arxiv.org)

FreshMem: Brain-Inspired Frequency-Space Hybrid Memory for Streaming Video Understanding (arxiv.org)

Bi-temporal Image-driven Acute Stroke Evolution Analysis (arxiv.org)

Speedbumps: Rejection Attacks on Speculative Decoding (arxiv.org)

Cross-Provider Review as a Runtime Contract for Coding Agents: A Controlled Pilot and Fault-Injection Study (arxiv.org)

iAm.md: Robot Skill Self-Assessment through Agentic Introspection for Unknown Open-Vocabulary Domains (arxiv.org)

Probabilistic Sensing, Deterministic Authority: Admitting Model-Produced Observations into Sufficiency-Checked Governance Contracts (arxiv.org)

Can a System-One LLM Perform Knowledge Tracing When Few or No Learners Are Logged? (arxiv.org)

Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation (arxiv.org)

The Missing Fourth Term for the Emulation Tensor Memory Equilibrium (TME) Model: The Residue Deconstruction Cost (arxiv.org)

Missing Modality-Aware Calibration for Trustworthy Brain Tumor Segmentation (arxiv.org)

State Stream Transformer (SST) V2: Parallel Training of Nonlinear Recurrence for Latent Space Reasoning (arxiv.org)

ProtoSemImage: Image-Valued Prototypes with Deformable Row Alignment for Interpretable Document Classification (arxiv.org)

GROB: A Multi-Agent Architecture for Public-Trace Investigation of Candidate Agentic Activity (arxiv.org)

RoboAware: Learning to Coordinate Embodied Skills from Counterfactual Outcomes (arxiv.org)

Large Language Model Turnover Undermines Screening for Artificial Intelligence-Assisted Scientific Writing (arxiv.org)

Beyond Report Imitation: Clinically Aware Multi-Image Ultrasound Report Generation from Visible Evidence (arxiv.org)

TAM: Task-Aware Memory Distillation for Efficient Spatiotemporal Prediction (arxiv.org)

Neural Networks for Temporal Pattern Recognition and Dynamic Arm Gesture Speed Estimation for Robot Control (arxiv.org)

LTBD: Learnable Trust-Boundary Delimiters for Prompt Injection Defense (arxiv.org)

Intent Graph: Navigating the Analytical Reasoning Space for Exploratory Data Analysis (arxiv.org)

Language Modeling is Monotone Compression (arxiv.org)

AffordDrive3D: Affordance-Aware World-Action Modeling with Spatial Understanding (arxiv.org)

Continuous Ground-Truth Construction and a Recovery Policy for Air--Water Robotic Tracking (arxiv.org)

CARE: Constrained Attention Refinement for Fine-Grained Visual Classification via Teacher-Student Distillation (arxiv.org)

Characterizing Statistical Separability in TP-CRIV for Probabilistic AI Models (arxiv.org)

PMTRM: Pseudo-Memory Temporal Re-encoding Module for Embodied Policy Learning (arxiv.org)

VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video Understanding (arxiv.org)

Defensive Sufficiency in a Stackelberg Model of AI Security (arxiv.org)

Higher-Order Action Supervision Makes A Strong Policy Class (arxiv.org)