The paper introduces a model‑free continuous‑time reinforcement learning algorithm that learns from a fixed batch of discrete‑time off‑policy data. It builds on the PhiBE‑Q framework to create a PhiBE‑PI policy iteration that uses a single state‑action critic, avoiding explicit drift or diffusion estimation. The authors provide a local end‑to‑end suboptimality guarantee under linear function approximation, analyze error propagation from discretization, function approximation, and finite samples, and reveal a bias‑variance trade‑off where finer time discretization can worsen statistical error.
By Chaorui Yao, Yuming Paul Zhang, Lin Yang, Yuhua Zhu
Limiting‑Kernel Q(λ) (LKQL) is an off‑policy value estimator that blends n‑step truncation with a long‑horizon approximation based on the limiting kernel. It maintains the computational efficiency of n‑step methods while improving policy evaluation accuracy, especially for long‑horizon tasks. The authors prove faster convergence of LKQL’s operator under aperiodicity and near‑on‑policy conditions, and demonstrate empirical gains on MuJoCo continuous‑control benchmarks.
By Tolga Ok, Arman Sharifi Kolarijani, Peyman Mohajerin Esfahani, Mohamad Amin Sharifi Kolarijani
The paper identifies an informational curse of horizon in goal-conditioned policy learning, showing that longer goal relabeling horizons during training can hurt policy generalization and performance. Controlled experiments with oracle planners separate training horizons from test horizons, revealing that behavioral cloning policies degrade with longer horizons while reinforcement learning mitigates this effect. The authors attribute this to reduced conditional mutual information between actions and hindsight-relabeled goals, and demonstrate that distilling short-horizon policy Jacobians into long-horizon policies improves performance, especially in combinatorial manipulation tasks.
By John L. Zhou, Yuxuan Dong, Jonathan C. Kao
arXiv:2606. 09825v1 Announce Type: cross Abstract: Training reinforcement learning (RL) policies from scratch is costly: it requires careful reward and environment design, extensive tuning, and substantial computation.
By Anton Bolychev, Georgiy Malaniya, Sinan Ibrahim, Pavel Osinenko
arXiv:2510. 03494v2 Announce Type: replace Abstract: We study finite-horizon offline reinforcement learning (RL) with function approximation for both policy evaluation and policy optimization.
By Volodymyr Tkachuk, Csaba Szepesv\'ari, Xiaoqi Tan
arXiv:2512. 14617v2 Announce Type: replace-cross Abstract: Many practical decision-making problems involve tasks whose success depends on the entire system history, rather than on achieving a state with desired properties.
By Alessandro Trapasso, Luca Iocchi, Fabio Patrizi
In value-based reinforcement learning, improving the accuracy of policy evaluation has been shown to improve downstream policy optimization performance. The widely adopted family of approximations rel...
The paper introduces a framework for reinforcement learning in Markov decision processes where policies are ordered by the information they provide, allowing data from one policy to estimate the performance of others. An epoch‑based algorithm is developed for infinite‑horizon average‑cost MDPs with continuous state and action spaces, achieving regret that scales with the width of the policy class’s information order. The authors apply the framework to lost‑sales inventory, dual‑sourcing, and state‑dependent queueing systems, demonstrating improved regret bounds and empirical performance without requiring global convexity or specific arrival‑rate structures.
By Zhongjun Zhang, Shipra Agrawal, Ilan Lobel, Sean R. Sinclair, Christina Lee Yu
The paper introduces state abstractions that preserve the difference of Q‑functions for offline reinforcement learning, aiming to exclude irrelevant dynamics from rich state data. It proposes a dynamic generalization of the R‑learner that uses orthogonal estimation and sparse learning to estimate the Q‑function contrast, achieving faster convergence and consistency under a margin condition. Experiments on simulated and simulator‑augmented real data show variance reductions and demonstrate that the necessary information for sequential decision‑making can be smaller than that required for full state prediction.
By Defu Cao, Angela Zhou
arXiv:2511. 19849v2 Announce Type: replace-cross Abstract: Recurrence objectives, where a target region must be visited infinitely often, are a fundamental class of specifications for Markov decision processes (MDPs) and form the core of $\omega$-regular and linear temporal logic (LTL) objectives.
By Dominik Wagner, Leon Witzman, Luke Ong
The paper introduces an asynchronous value‑iteration RL algorithm for average‑reward multichain MDPs that requires only the transition graph and uses Bather’s decomposition to partition the state space into communicating subsystems and transient states. This hierarchical decomposition transforms the global decision problem into structured subproblems, enabling the algorithm to converge to the optimal gain and produce gain‑optimal policies in finite time. Two additional algorithms are presented: one that approximately solves the multichain average optimality equations for near gain‑optimality, and another that approximates the optimal bias function to achieve near bias‑optimality, both with almost‑sure convergence guarantees and improved transient performance.
whyItMatters":"The work offers the first essentially model‑free average‑reward RL algorithms for general multichain MDPs, avoiding reductions to discounted problems."
By Huizhen Yu, Isaiah Heidt
arXiv:2501.06926v5 Announce Type: replace
Abstract: Double reinforcement learning (DRL) provides efficient off-policy inference for policy values in nonparametric Markov decision processes (MDPs), bu...
By Lars van der Laan, David Hubbard, Allen Tran, Nathan Kallus, Aur\'{e}lien Bibaut