AI News

Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning (arxiv.org)

Decision Transformer for UAV-Mounted RIS-Assisted Dynamic D2D Communications (arxiv.org)

Legible Failures: Detecting and Repairing In-Context Binding Errors (arxiv.org)

Scored vs. Generated Readouts in Behavioral Language Models: An Empirical Study of Elicitation Format (arxiv.org)

DNA: Differentially private Neural Augmentation for contact tracing (arxiv.org)

Musec: MomentUm SpEctral Clipping for Stable Muon-type Training (arxiv.org)

ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding (arxiv.org)

Reliable Near-Field Multi-User Positioning Informed by Two-Stage MUSIC (arxiv.org)

Phase-Decoupled, Model-Calibrated Power Control for Disaggregated LLM Serving (arxiv.org)

Beyond Training: A Feasibility Taxonomy for Inference-Time AI Governance (arxiv.org)

Quantifying Logical Consistency in Transformers via Query-Key Alignment (arxiv.org)

Building Supervision into Hebbian Plasticity through Spike Agreement (arxiv.org)

Addressing A Posteriori Performance Degradation in Neural Network Subgrid Stress Models (arxiv.org)

No Screening is More Efficient with Multiple Objects (arxiv.org)

When do cheap embeddings beat protein language models? A theoretically-grounded hashing sketch for biological sequence classification (arxiv.org)

scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning (arxiv.org)

How Wrong Can a Good Predictor Be? Diverging Updates with Vanishing Predictive KL (arxiv.org)

MUtE: A Dual Framework for Concept Erasure and Counterfactual Interventions (arxiv.org)

Deep operator learning for efficient sampling from invariant measures of stochastic differential equations (arxiv.org)

Predicting Train Delays in Finland Using Machine Learning and Weather Data (arxiv.org)

Project Qualia: Recovering Experiential Music Structure from Session Co-occurrence Data (arxiv.org)

SynCo: Synthetic Community-Aware Attributed Graph Generator for Graph Neural Network Benchmarking (arxiv.org)

Beyond Solver Verdicts: Generative Reward Models for Autoformalization (arxiv.org)

TART: A Modular Tool for Technique-Aware Audio-to-Tablature Guitar Transcription (arxiv.org)

LogiScope-VQA: Benchmarking Vision-Language Models for Logistics Hazard Identification in Industrial Scenarios (arxiv.org)

Context operations to architecture modelling output from large language models and evaluation criteria for their use in systems engineering design (arxiv.org)

Semi-Tensor Product-Based Multi-Term Randomized T-SVD and Its Visual Applications (arxiv.org)

The information geometry of large language models is shared, learned, and controllable (arxiv.org)

Why Sample What You Can Enumerate? Exact Policy Optimization for Genomic Tool Selection (arxiv.org)

Solving Few-Shot Multiobjective Multitask Optimization via Iterative Sequential Transfer (arxiv.org)

Optimizing Three Critical Factors for Practical and Effective OOD Detection Fine-Tuning (arxiv.org)

When Noise Fabricates Bias: The Fragility of LLM-as-a-Judge Bias Measurement under Noisy Text (arxiv.org)

EGGROLL, Unrolled: Understanding and Improving Low-Rank Evolution Strategies at Scale (arxiv.org)

What Should an Agent Forget? Separating What Is Stored from What Is Used (arxiv.org)

Understanding LoRA Rank Trade-offs in Diffusion Model Fine-Tuning (arxiv.org)

Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to Repeated Data (arxiv.org)

Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation (arxiv.org)

DriftNet: A Dual-Head Trajectory Transformer for Detecting and Localizing Prompt Injection in LLM Agents (arxiv.org)

Reflex-Informed Neuromuscular Reinforcement Learning for Muscle-Driven Locomotion (arxiv.org)

From Protocols to Evidence: Bounded Claims for AI in Service of the Common Good (arxiv.org)

Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation (arxiv.org)

Polyhedral Geometry of Time-to-First-Spike Neural Networks (arxiv.org)

An Experimental Evaluation of Multimodal Prompt Injection Attacks on Agentic AI Frameworks (arxiv.org)

The Era by Eon Benchmark: A Generated Enterprise Estate with Exact Ground Truth for Benchmarking LLM Agents (arxiv.org)

Can Artificial Intelligence Support Healthcare and Mental Health Through Early Cyberbullying Detection ? The Impact of Emotion-Aware AI on Proactive Online Safety (arxiv.org)

When does a spectral prior help graph learning? Connectivity-loss estimation under road-network disruptions (arxiv.org)

Grounded Evaluation and Repair for NL-to-PDDL Problem Generation (arxiv.org)

UnitBoost: Managing Compound LLM Systems with a Merge Operator, Not a Model (arxiv.org)

Sparsity Regularized and Robust Mean Variance Portfolio Selection Under Ellipsoidal Uncertainty (arxiv.org)

CS-Guard: Benchmarking LLM Guardrails for Code Generation Security (arxiv.org)