AI News

Backward SDEs-based Diffusion for Physics-Constrained Generation (arxiv.org)

FedLTLib: A Comprehensive Benchmark for Federated Long-Tail Learning (arxiv.org)

Where to Compute and How to Interact: Operator-Readable Adaptation with Gauge-Aware Transport (arxiv.org)

Multi-View Molecular Representation Learning with Hierarchical Graphs and Contextualized Fingerprints (arxiv.org)

Bayesian Optimisation Using Product-of-Experts Gaussian Process Models with Uncertainty Calibration (arxiv.org)

The Token Before the Value Is the Key: How Hybrid Architectures Organize Induction Circuits (arxiv.org)

Principal-timestep Restricted Init via Sparse Matrix-decomposition in Flow-matching (arxiv.org)

Specifying Reward Functions for RL Without Environment Sampling (arxiv.org)

The Misery of Mechanistic Interpretability: A Formal Perspective (arxiv.org)

Beyond Noise: Understanding and Overcoming Temperature Effects in Analog DNN Inference (arxiv.org)

End-to-End Verifiable and Robust Federated Learning (arxiv.org)

Same path, different: a mechanistic comparison of looped and stacked transformer encoders on 12-lead ECG (arxiv.org)

Rotation-Based Subspace Tracking for Robust Kernel PCA on Streaming Data (arxiv.org)

Data-driven Prediction of Satellite-observed Avalanche Activity from Snowpack Simulations (arxiv.org)

Ensemble Complexity in Photovoltaic Forecasting (arxiv.org)

Privacy Preserving Gossip Learning (arxiv.org)

WaVeFuse: Regime-Adaptive Equity Index Forecasting via Channel-Wise Wavelet Denoising and Vertical Attention Fusion (arxiv.org)

Are Gradient Boosting Models Suitable for Intermittent Demand Forecasting? (arxiv.org)

Carryover Drafting: Recycling Rejected States for Speculative Decoding (arxiv.org)

GRPO-QM: Target Preserving Exploration for Quantum Tomography (arxiv.org)

An immune world model for multiscale forecasting and therapeutic hypothesis generation (arxiv.org)

SH-WRNN: Implicit Spherical Harmonics Weight Field Routing Neural Networks for Asymmetric Edge Intelligence (arxiv.org)

Diagnosing Temporal Misalignment in Multichannel Time-Series Classification with Minimum Description Length (arxiv.org)

Pathwise Individual Rationality in Federated Learning: A Mechanism-Architecture Co-Design (arxiv.org)

Selecting k Paths with the Minimum Longest Path Length in the Stochastic Semi-Bandit Setting (arxiv.org)

GNN4PPM: Multi-Target Predictive Process Monitoring with Relational Graph Convolutional Networks (arxiv.org)

Should All Noises Be Treated Equally: Impact of Input Noise Variability on Neural Network Robustness (arxiv.org)

EdgeHAR: An Edge-Native Compact Sensor Foundation Model for Human Activity Recognition (arxiv.org)

Retrieval-Guided Fine-Tuning as Noisy Estimation: Risk bounds and Architectural Analysis (arxiv.org)

Follow the Geometry, Not the Model: Cold Start Semi-Supervised Learning (arxiv.org)

Towards Identifying the Dataset Biases Causing Phantom Transfer (arxiv.org)

Multimodal deep learning from spectra for small-molecule structure identification: enhancing robustness with mixed-condition training (arxiv.org)

Communication-Efficient LLM Adaptation over Decentralized GPU Meshes (arxiv.org)

Hardware-Aware Learned Representation Compression for Distributed In-Sensor Vision (arxiv.org)

Mitigating the Stability-Plasticity Dilemma in Adaptive Train Scheduling with Curriculum-Driven Continual DQN Expansion (arxiv.org)

Disentangling Representation Evolution in Transformers through Directional Decomposition (arxiv.org)

Mind2Dialogue: Training Human-Aware Language Models by Simulating User Mental States (arxiv.org)

Know When to Stop, Where to Restart: Accelerating Multi-Turn Agentic On-Policy Distillation (arxiv.org)

Adaptive Phase-Switching for Communication-Efficient Federated LoRA Fine-Tuning (arxiv.org)

A Three-Axis Stress Test of LLM vs Classical ML for Network Intrusion Detection under Distribution Shift and Adversarial Evasion (arxiv.org)

Pretraining for Sample-Efficient Neural Interfaces (arxiv.org)

One Spectrum, Two Resources: Data-Memory Scaling in Autoregressive Prediction (arxiv.org)

On the Potential of Multi-Task Learning in Predictive Process Monitoring (arxiv.org)

A Machine Learning API for Earth Observation Data Cubes Based on openEO (arxiv.org)

Learning to Solve Hard Problems in RL for LLMs by Never Giving Up (arxiv.org)

Efficient Online Inverse Optimization with $O(d)$ Regret (arxiv.org)

Optimal Transport for Efficient, Unsupervised Anomaly Detection on Industrial Data (arxiv.org)

Exploring napping paradigm for Recurrent Spiking Neural Networks (arxiv.org)

Minibatch persistency, eight years later: what batch reuse costs in steps and joules, and what it saves in data (arxiv.org)

AgentKV: Phase-Aware KV Eviction for Agentic LLMs (arxiv.org)