AI News

MoEZip: Routing-Aware KV Cache Compression for Sparse Mixture-of-Experts LLMs (arxiv.org)

Random Remasking Scales Masked Diffusion Inference (arxiv.org)

PHD-Hand: A Printable Humanoid Dexterous Hand with Linkage-Driven 3-DOA Digits (arxiv.org)

Building Auditable AI Text Detectors with Sparse Autoencoders (arxiv.org)

Every Alarm Has a Cause: An Open Generator of Telecom Alarm Storms with Exact Root Causes, Bound to NORIA-O, RFC 8632 and TMF642 (arxiv.org)

HybridResearcher: Bridging the Structured-Unstructured Boundary for Deep Research Agents (arxiv.org)

TerraTransfer++: Demonstration-Free End-to-End Planner (arxiv.org)

AQL-Lite: Enabling Efficient Knowledge Acquisition for Arabic LLMs via Data-Centric Continual Pretraining (arxiv.org)

An Exact 93-Move Solution to a Nonlocal Crystal-Exchange Puzzle (arxiv.org)

SciFusion: Physics-Guided Compositional Diffusion for Multimodal 3D Scientific Image Generation (arxiv.org)

Graph Topology Meets Energy-Driven Pathfinding: Preserving Contextual Consistency in Retrieval-Augmented Generation (arxiv.org)

EgoHand: Monocular 3D Hand Reconstruction from Egocentric Video (arxiv.org)

HiddenLens: A Multi-Layer LLM Framework for Detecting and Resolving Hidden Assumptions in Software Requirements (arxiv.org)

Post-Hoc Feature Selection Layer for Neural Networks Interpretability (arxiv.org)

On the Token Collapse Phenomenon in Textual Graph Soft Prompting (arxiv.org)

PAUSE: Search-Free Steering of Coding Agents Before Their First Source-Code Edit (arxiv.org)

FaultGene: LLM-Assisted Fault-Augmented Test Generation for Hardware-in-the-Loop Validation of Autonomous Driving Systems (arxiv.org)

Every Part Has Its Role: Part-of-Speech Effects in LLM Post-Training (arxiv.org)

Automated Materials Data and Knowledge Extraction using Machine Learning and Large Language Models (arxiv.org)

Adaptive Deadline- and Energy-Aware Caching Algorithms in Dynamic Multi-UAV Edge Networks (arxiv.org)

Keeping Time vs. Understanding Time: Neither Alone Explains Temporal Reasoning in LLMs (arxiv.org)

nnInteractive Agent: General-Purpose AI Agents as 3D Image Annotators (arxiv.org)

Is High Local Accuracy Enough? Geometric Consensus with Support-Adaptive Statistical Personalization for Federated Learning (arxiv.org)

Metamodel-Guided LLM Decoding for Consistent DSL Generation (arxiv.org)

Robot Generalization Beyond Volume: The Diversity Path via Zero-Shot Capabilities (arxiv.org)

Ground-Truth-First Benchmarking of Reinforcement Learning via Converse Optimality: From Smooth, Fully Observed Systems to Non-Smooth or Partially Observed Tasks (arxiv.org)

Tracing Subliminal Learning Through Pretraining (arxiv.org)

Retrieval Augmented Generation Compositionality (arxiv.org)

SAKURA: Serving Attention Key-Value State Under Rate Adaptation for Private Image Editing on Phones (arxiv.org)

S-EDL: Eliciting Self-Evidence from Sequence Likelihoods for Semantic Calibration of LLMs (arxiv.org)

Joules-to-GDP: How Much Economic Value Do AI Systems Produce per Unit of Energy? (arxiv.org)

An interpretable deep learning method for early hepatocellular carcinoma diagnosis on multiphase CT (arxiv.org)

One Pretrained Decision Model Is Enough for Training-Free Generalist Graph Anomaly Detection (arxiv.org)

Conflicted Records: Typed Adversarial Contracts for Auditable LLM Output Quality Enforcement (arxiv.org)

FogTTA: Online Test-Time Adaptation for Robust Transformer-based Object Detection in Foggy Weather (arxiv.org)

GIFT: Gradient Informed Filtering for Distillation Augmented Reasoning RL (arxiv.org)

Primal-Dual Representation Learning for Low-Rank Constrained MDPs (arxiv.org)

Zero-Cost Detection of Fabricated Citations: A Staged Heuristic Approach for Computing Dissertation Assessment (arxiv.org)

Formalizing Cooperation and Competition: An AI-Assisted Mathematical Addendum to a 2018 Planning Thesis (arxiv.org)

Trace2World: Reconstructing App Worlds from Real Interaction Traces for Scalable GUI Agent Training (arxiv.org)

Sensorimotor Alignment: A Platonic Proxy for Embodied Policy Success (arxiv.org)

Overview of CHIP 2026 Shared Task 1: Patient-Specific Phenotype Concept Recognition in Full-Text Biomedical Literature (arxiv.org)

CulturalRiddles: A Multicultural Benchmark of Riddles (arxiv.org)

GRAFT: Restoring the Missing Sampling Interface in Continuous Diffusion Language Models (arxiv.org)

Bio-Inspired Auditory Processing: Coupling Parallel Resonate-and-Fire Dynamics with Ternary Recurrent Networks for Speech Recognition (arxiv.org)

Visualizing and Interpreting Temporal Care Team Dynamics for Cancer Survival Risk Analysis (arxiv.org)

From AI-Free Gates to AI-First Work: How Engineering Students in India Navigate University Education and Campus Placements Without AI Guidelines (arxiv.org)

DiffLTM: Continuous Diffusions Large Tabular Models for Conditional Generation (arxiv.org)

Rethinking SAE Evaluation: An Atomic Interpretable Unit Framework Reveals Hidden Polysemanticity and Redundancy (arxiv.org)

Wavenumber-Aware Deep Learning for Robust Analysis of Raman Spectra (arxiv.org)