←── back to feed
/topics/arxiv-machine-learning-papers-september-25

arXiv machine learning papers September 25

26 items●1 sources●updated 44d ago●trend 0

On September 25, 2026, arXiv released 20 machine learning papers spanning reinforcement learning, diffusion models, probabilistic inference, neural operators, and anomaly detection. Key contributions include methods for step-level advantage estimation in agentic RL, fully-Bayesian guided diffusion for inverse problems, neural transport sampling for molecular systems, and theoretical results on time-series validation trade-offs.

  • RLVR landscapes paper (arXiv:2609.28625) maps entropy-regularized RL to spin-glass models to characterize optimization landscape on algorithmic tasks
  • GRPO step-level advantage estimation (arXiv:2609.28963) addresses systematic bias in group-normalized RL for reasoning LLMs using trajectory graphs
  • FB-GDM (arXiv:2609.29216) removes hyperparameter tuning for diffusion-based inverse problems via fully-Bayesian guided approach
  • Neural Transport Nested Sampling (arXiv:2609.29413) combines nested sampling with flow-matching for Boltzmann distribution sampling in molecular systems
  • Time-series validation impossibility theorem (arXiv:2609.29530) proves three validation criteria—training sufficiency, test coverage, temporal causality—cannot be simultaneously satisfied
[BLG]blog/rss26
RLVR landscapes for iterated multiplications can be benign: Insights from spin-glass theory
arXiv stat.ML · Noa Rubin, Zohar Ringel · 44d
Back to the Definition: Estimating Step-Level Advantages via Trajectory Graphs for Agentic Reinforcement Learning
arXiv stat.ML · Xincheng Yao, Haobo Fu, Weiming Liu, Chongyang Zhang · 44d
When Does Action Credit Need Updating?
arXiv stat.ML · Hongye Yang, Boxiao Huang · 44d
Functional dynamic mode decomposition: Learning infinite-dimensional systems from data
arXiv stat.ML · Stefan Klus, Eirini Ioannou · 44d
FB-GDM: Fully-Bayesian Guided Diffusion Models for High-Dimensional Linear Inverse Problems via Unsupervised Variational Inference
arXiv stat.ML · Gatien S\'eguy (SATIE), Thomas Rodet (SATIE) · 44d
Sufficiently Reduced Distributional Regression
arXiv stat.ML · Alexander Henzi, Tiange Liu, Xinwei Shen · 44d
Learning a Flow to Self-Supervised Representations
arXiv stat.ML · Yuling Jiao, Wensen Ma, Houduo Qi, Defeng Sun · 44d
Neural Transport Nested Sampling
arXiv stat.ML · David Yallup, Will Handley · 44d
Direct Message Approximation (DMA): A Consistency-Based Framework for Tractable Approximate Inference on Factor Graphs
arXiv stat.ML · Ralf Herbrich, Rainer Schlosser, Jan Lemcke, Johann Ukrow, Anna Kazachkova, Nicolas Alder, Leonhard Hennicke, Theo Bardey, Nico Grimm, Luca Kleinschmidt, Philipp Kolbe, Cezary Kujath, Johanna Schlimme, Karl Matti Sch\"utz · 44d
Physics-Informed Neural Operator Surrogate for 2D Magnetohydrodynamic Reconnection
arXiv stat.ML · Kushaal Kumar Pothula, Arunav Kumar · 44d
The Impossible Trinity of Time-Series Validation: A Conservation Law among Training Sufficiency, Test Coverage, and Temporal Causality
arXiv stat.ML · Jiayu Li · 44d
DeepGOF-1: A Pretrained Convolutional Goodness-of-Fit Test for Logistic Regression with a Computable Consistency Certificate
arXiv stat.ML · Ebrahim Khaled Ebrahim · 44d
When Identical Rows Disagree: From Benchmark Identifiability to Replication-Robust Anomaly Detection
arXiv stat.ML · Jie Deng · 44d
Stochastic Semantic Evidence Graphs: Uncertainty Propagation and Governance for Agentic AI
arXiv stat.ML · Matthew Francis Dixon · 44d
Boolean threshold functions, neuron capacity, and memory retrieval
arXiv stat.ML · Xinyuan Xie · 44d
Shrinking-Tube Concentration for Adaptive Markovian Stochastic Approximation
arXiv stat.ML · Jin Li, Ye Luo, Xiaowei Zhang · 44d
Intrinsic-Extrinsic Coupling in Learning Dynamics
arXiv stat.ML · Qinyou Wang · 44d
Anchored Extra-Proximal Methods: Optimal Higher-Order Methods for Monotone Inclusion Problems
arXiv stat.ML · Ruichen Jiang, TaeHo Yoon · 44d
Riemannian Gradient Descent for Gaussian Mixture Models with unknown diagonal covariances
arXiv stat.ML · Romane Giard, Yohann De Castro, Roland Denis, Cl\'ement Marteau · 44d
Deep Positive-Unlabeled Anomaly Detection for Contaminated Unlabeled Data
arXiv stat.ML · Hiroshi Takahashi, Tomoharu Iwata, Atsutoshi Kumagai, Yuuki Yamanaka · 44d
Time-Varying Bayesian Optimization Without a Metronome
arXiv stat.ML · Anthony Bardou, Patrick Thiran · 44d
Stacked SVD or SVD stacked? A Random Matrix Theory perspective on data integration
arXiv stat.ML · Tavor Z. Baharav, Phillip B. Nicol, Rafael A. Irizarry, Rong Ma · 44d
MultiwayPAM: Multiway Partitioning Around Medoids for LLM-as-a-Judge Score Analysis
arXiv stat.ML · Chihiro Watanabe, Jingyu Sun · 44d
CatSIM: A Categorical Image Similarity Metric
arXiv stat.ML · Geoffrey Z. Thompson, Ranjan Maitra · 44d
A Probabilistic Approach for Model Alignment with Human Comparisons
arXiv stat.ML · Junyu Cao, Mohsen Bayati · 44d
Capturing Unseen Spatial Heat Extremes Through Dependence-Aware Generative Modeling
arXiv stat.ML · Xinyue Liu, Xiao Peng, Shuyue Yan, Yuntian Chen, Dongxiao Zhang, Zhixiao Niu, Hui-Min Wang, Xiaogang He · 44d