AI News

GraFT: A Training-Free Framework for Spatial Reasoning in Multimodal Large Language Models via 3D Scene Graphs (arxiv.org)

FWBC-VLA: Force-Aware Whole-Body Compensation for Contact-Rich Loco-Manipulation (arxiv.org)

A Blind Trust, the Bloody Thrust: When Attacker-Controlled Hook Updates Steer AI Agent Harnesses towards Malicious Behaviors (arxiv.org)

GazeFS: Target-Centered Gaze-Trajectory Forecasting and Stabilization from Gaze-Head History (arxiv.org)

The impact of phase information for few-shot fine-grained image classification (arxiv.org)

LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes (arxiv.org)

IndicSafeEval: Safety Robustness of Large Language Models under Multilingual Persuasive Jailbreak Attacks (arxiv.org)

Data-efficient Kernel Methods for Learning Hamiltonian Systems (arxiv.org)

ENEAS: Embedding-guided Neural Ensemble for Adaptive Segmentation (arxiv.org)

Beyond BLEU: A Case for Redefining Sign Language Translation Benchmarks (arxiv.org)

Can LLMs Extract Architectural Design Decisions from Source Code Commits? - A Preliminary Exploratory Study (arxiv.org)

Symmetries and Causality: Causal Effect Identification Beyond IID Data (arxiv.org)

Cross-Dataset Transfer and Reliability of Explainable Artificial Intelligence for RhythmFormer Remote Photoplethysmography (arxiv.org)

Enhancing Financial Question Answering: A Novel Benchmark Dataset of Banks' financial statements (arxiv.org)

Doesn't Stop Reasoning: Analysis of Spurious CoT Termination (arxiv.org)

EraseSAE: Surgical Concept Erasure in Text-to-Video Diffusion Models via Sparse Autoencoders (arxiv.org)

Test-time adaptation for speech enhancement with an autoregressive speech prior (arxiv.org)

ToolDF: Tool-Integrated Reasoning for Mixed-Authenticity Audio Deepfake Detection (arxiv.org)

Remember and Reweight: Enhancing Multi-Agent Debate with Experience Memory and Confidence Estimation (arxiv.org)

FailBench: How Reliable are VLMs at Judging Robot Task Success? (arxiv.org)

How Far Can Synthetic Data Take Thai OCR? (arxiv.org)

LevelSyn: Physical-Aware Logic Synthesis via Level-Asynchronous Graph Neural Networks (arxiv.org)

From Prior-Guided Heuristics to Deployable Agents: Accelerating Demonstration-Driven Reinforcement Learning for Deadline-Constrained Network Control (arxiv.org)

Toward Physically Grounded JEPA World Models for Goal-Conditioned Robotic Planning (arxiv.org)

WIDE: Wildcard Inference with Dynamic Expansion for Cross-Modal Generative Retrieval (arxiv.org)

FedPS: Federated Preprocessing for structured data via aggregated Statistics (arxiv.org)

Temperature Scaling Attack Disrupting Model Confidence in Federated Learning (arxiv.org)

MSign: An Optimizer Preventing Training Instability in Large Language Models via Stable Rank Restoration (arxiv.org)

Linearized subspace refinement framework to expose hidden accuracy in trained neural networks (arxiv.org)

DuaDeep-SeqAffinity: Dual-Branch Deep Learning for Tri-Stream Sequence-Based Antibody--Antigen Affinity Prediction (arxiv.org)

Attention Trajectories as a Diagnostic Axis for Deep Reinforcement Learning (arxiv.org)

Adaptive Resolving Methods for Markov Decision Processes with Function Approximations (arxiv.org)

Learning Constraints-Based Adaptive Hypergraph Neural Networks for Solving Vehicle Routing Problems (arxiv.org)

A cautionary tale on the cost-effectiveness of collaborative AI in real-world medical applications (arxiv.org)

Anisotropic View Distance Metric for High-Dimensional Data: Theory, Geometry, and Fast Computation (arxiv.org)

Compile by Training: Turning Natural-Language Specifications into Local Neural Functions (arxiv.org)

Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning (arxiv.org)

Towards Universal Tabular Embeddings: A Benchmark Across Data Tasks (arxiv.org)

Para-Pipe: Exploiting Hierarchical Operator Parallelism of ML Computational Graphs on SoCs (arxiv.org)

The Head Complexity of Boolean Functions in Single-Layer Attention (arxiv.org)

Influence of Extruded Filament Shape on Buildability in 3D Concrete Printing: A Geometry-Informed Deep Learning-FEM Approach (arxiv.org)

Differentiable Hybrid Modelling for Learning and Optimising Chemical Transport Processes from Experimental Data (arxiv.org)

Sharpening the Ensemble: An SSIM-Aligned Residual Refiner for Brain-MRI Inpainting Post-Processing (arxiv.org)

Parameterised graph theory for tensor networks: entanglement rerouting, structural simplification, and agnostic tomography (arxiv.org)

RARF: Region-Aware Rectified Flows for 3D Brain MRI Inpainting (arxiv.org)

Seeing Before Synthesizing: VLM-Guided Transition Event Discovery for Weakly-Supervised Dense Video Captioning (arxiv.org)

TruncGradGS: Improved 3D Gaussian Splatting via Truncated Gradient Updates (arxiv.org)

The Civilization Framework: Sovereign-Anchored Communication Between Personal Multi-Agent Systems (arxiv.org)

Privacy, Robustness, and Fairness Trade-offs in Federated Intrusion Detection: Geometric Indistinguishability at the Aggregation Interface (arxiv.org)

StrixAE: An Intelligent Agent for Audio Enhancement under Complex Distortion Coupling in Real-World Scenarios (arxiv.org)