arXiv · cs.LG
Recent machine learning papers from arXiv.
Read from its RSS feedAppears in:IA
Latest from arXiv · cs.LG
- Evolving to Match Capabilities: Improving Harness Self-Evolution via Skill Adaptationhá 11 d
- Can Language Models Learn While They Think? Testing Gradient Adaptation Across Architectureshá 11 d
- Lessons from Luck: Do LLM-Written Lessons Carry Chance into an Agent's Next Decision?há 11 d
- Feature Scaling Beats Architecture Choice in Offline Respiratory Screening for Low-Resource Clinicshá 11 d
- Inference-Time Prior Adaptation for Imbalanced Data Streams under Concept Drifthá 11 d
- Physics Informed Neural Networks for Solving the Drift-Diffusion Equationshá 11 d
- Neural-DISCO: Source-Conditioned Counterfactual Editing of Neural Population Activityhá 11 d
- How Much Fresh Data Should CTR Recalibration Use? A Temporal Evaluation of Post-hoc Calibrationhá 11 d
- Lower expected Fisher rank implies better PAC generalizationhá 11 d
- Model-agnostic predictability quantification of multivariable time serieshá 11 d
- Vision Token Pruning via Query-Vision Interaction Decompositionhá 11 d
- Neyman Orthogonal Activation Steeringhá 11 d
- Reinforcement Learning for Controller Parameters in the Auto-Bidding System at Pinteresthá 11 d
- RadonBench: A Benchmark of Classical and Learning-Guided Search Techniques for Problems in Combinatorial Geometryhá 11 d
- Individual Fairness via Local Predictive Consistencyhá 11 d
- All-in-One: A Composable Attack on LLM Transformationshá 11 d
- Tree Depth and Sample-Quality-Associated Features: Two Hidden Factors in Boruta Selection for RNA-Seq Data Classificationhá 11 d
- When Does Pre-Trained Capacity Help Causal Learning? From Scalar Prediction to Counterfactual Post-Traininghá 11 d
- MLIPDiff: Guiding Denoising Diffusion Models for Structure-Based Drug Design via Machine-Learned Interatomic Potentialshá 11 d
- Augmentation-Guided Adversarial Training via Local Stability-Based Sample Selectionhá 11 d
- Gaussian Approximation for Composition and Subsampling of Differentially Private Algorithmshá 11 d
- Information-Theoretic Scaling Laws for Generative Diffusion Traininghá 11 d
- The Clustering Occupation Kernel (CLOCK) Method for Time Series Arising from Dynamical Systemshá 11 d
- LSTM-Style Looped Transformers with Shared KV Cacheshá 11 d
- Multiple Importance Sampling for Guided Speculative Inferencehá 11 d
- One Call Ahead: Packed Self-Speculative Decoding with Diffusion Draftinghá 11 d
- The Cost of an Unknown Horizon for the Last Iterate of Clipped SGD under Heavy-Tailed Noisehá 11 d
- Faithful Federated Tabular Data Synthesishá 11 d
- COPE: Composable Performance Emulation for Mixture-of-Experts Offloading Policieshá 11 d
- Performance Gaps Between Open-Loop, Certainty-Equivalent, and Optimal Control Under Action-Dependent Observationshá 11 d