arXiv Machine Learning July 21, 2026 By Keru Wang, Yixin Deng, Yao Lyu, Stephen Redmond, Shengbo Eben Li

Distributional Soft Bellman Operator under the Cram\'er Geometry

Read the original on arXiv Machine Learning →

arXiv:2607. 17897v1 Announce Type: new Abstract: Distributional soft policy iteration (DSPI) provides an important framework for combining distributional reinforcement learning (DRL) with maximum-entropy control, in which the policy evaluation step is governed by a distributional soft Bellman operator acting on entropy-regularised returns.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv Machine Learning.

reinforcement-learning

arXiv Machine Learning

Jun 9

Global Convergence of Wasserstein Policy Gradient for Entropy-Regularized Reinforcement Learning

arXiv:2605. 26078v3 Announce Type: replace Abstract: Wasserstein policy gradient (WPG) is a policy optimization method for reinforcement learning (RL) that exploits the optimal-transport geometry of action distributions.

By Zhaoyu Zhu, Rui Gao, Shuang Li

diffusion reinforcement-learning safety

arXiv Machine Learning

Jun 16

Maximum Entropy Inverse Reinforcement Learning for Mean-Field Games with Average Reward

arXiv:2606. 16759v1 Announce Type: new Abstract: We study inverse reinforcement learning for discrete-time, infinite-horizon mean-field games (MFGs) under an average-reward criterion.

By \c{S}evket Kaan Alk{\i}r, Naci Sald{\i}, Berkay Anahtarc{\i}, Can Deha Kar{\i}ks{\i}z

reinforcement-learning

Hugging Face Trending Papers

Jul 6

Fitted Occupancy-Ratio Evaluation without Bellman Completeness

Occupancy ratios correct distribution shift in offline reinforcement learning and are central to off-policy evaluation. Existing primal-dual and minimax methods typically estimate these ratios by enforcing occupancy-balance moments over a critic class.

reinforcement-learning

arXiv AI

Jun 16

Direction-Conditioned Policies via Compositional Subgoal Scoring for Online Goal-Conditioned Reinforcement Learning

arXiv:2606. 16515v1 Announce Type: cross Abstract: Hamilton-Jacobi-Bellman theory implies that the optimal goal-conditioned action depends on the goal only through the gradient of the goal-reaching distance at the current state, yet standard online GCRL still conditions the actor on the raw goal -- a signal that is geometrically uninformative when the goal is far from the data distribution.

By Swaminathan S K, Damiya Gondha, Theyanesh Eswaramoorthy Rajahkrishnan, Aritra Hazra

reinforcement-learning robotics

arXiv Machine Learning

Jul 17

Bridging the Gap between Newton-Raphson Method and Regularized Policy Iteration

arXiv:2310. 07211v2 Announce Type: replace Abstract: Regularization is a cornerstone of modern reinforcement learning.

By Zeyang Li, Chuxiong Hu, Yunan Wang, Guojian Zhan, Jie Li, Yao Lyu, Shengbo Eben Li

reinforcement-learning

arXiv Machine Learning

Jul 7

Fitted Occupancy-Ratio Evaluation without Bellman Completeness

arXiv:2607. 05375v1 Announce Type: cross Abstract: Occupancy ratios correct distribution shift in offline reinforcement learning and are central to off-policy evaluation.

By Lars van der Laan, Nathan Kallus

reinforcement-learning

Related stories

Global Convergence of Wasserstein Policy Gradient for Entropy-Regularized Reinforcement Learning

Maximum Entropy Inverse Reinforcement Learning for Mean-Field Games with Average Reward

Fitted Occupancy-Ratio Evaluation without Bellman Completeness

Direction-Conditioned Policies via Compositional Subgoal Scoring for Online Goal-Conditioned Reinforcement Learning

Bridging the Gap between Newton-Raphson Method and Regularized Policy Iteration

Fitted Occupancy-Ratio Evaluation without Bellman Completeness