English
Related papers

Related papers: Fitted $Q$ Evaluation Without Bellman Completeness…

200 papers

In reinforcement learning, distributional off-policy evaluation (OPE) focuses on estimating the return distribution of a target policy using offline data collected under a different policy. This work focuses on extending the widely used…

Machine Learning · Statistics 2025-10-21 Sungee Hong , Jiayi Wang , Zhengling Qi , Raymond K. W. Wong

$Q$-learning is one of the most fundamental reinforcement learning algorithms. It is widely believed that $Q$-learning with linear function approximation (i.e., linear $Q$-learning) suffers from possible divergence until the recent work…

Machine Learning · Computer Science 2025-05-28 Xinyu Liu , Zixuan Xie , Shangtong Zhang

Many reinforcement learning algorithms rely on value estimation, however, the most widely used algorithms -- namely temporal difference algorithms -- can diverge under both off-policy sampling and nonlinear function approximation. Many…

Machine Learning · Computer Science 2024-08-02 Andrew Patterson , Adam White , Martha White

In recent work it is shown that Q-learning with linear function approximation is stable, in the sense of bounded parameter estimates, under the $(\varepsilon,\kappa)$-tamed Gibbs policy; $\kappa$ is inverse temperature, and $\varepsilon>0$…

Machine Learning · Computer Science 2026-02-09 Prashant Mehta , Sean Meyn

We study the off-policy evaluation (OPE) problem in reinforcement learning with linear function approximation, which aims to estimate the value function of a target policy based on the offline data collected by a behavior policy. We propose…

Machine Learning · Computer Science 2022-01-05 Yifei Min , Tianhao Wang , Dongruo Zhou , Quanquan Gu

We consider the off-policy evaluation problem of reinforcement learning using deep convolutional neural networks. We analyze the deep fitted Q-evaluation method for estimating the expected cumulative reward of a target policy, when the data…

Machine Learning · Computer Science 2022-10-05 Xiang Ji , Minshuo Chen , Mengdi Wang , Tuo Zhao

Effectively leveraging large, previously collected datasets in reinforcement learning (RL) is a key challenge for large-scale real-world applications. Offline RL algorithms promise to learn effective policies from previously-collected,…

Machine Learning · Computer Science 2020-08-20 Aviral Kumar , Aurick Zhou , George Tucker , Sergey Levine

Offline reinforcement learning is important in domains such as medicine, economics, and e-commerce where online experimentation is costly, dangerous or unethical, and where the true model is unknown. However, most methods assume all…

Machine Learning · Statistics 2025-10-30 David Bruns-Smith , Angela Zhou

Offline reinforcement learning (RL) seeks to learn optimal policies from static datasets without further environment interaction. A key challenge is the distribution shift between the learned and behavior policies, leading to…

Machine Learning · Computer Science 2025-08-11 Haohui Chen , Zhiyong Chen

Policy gradient (PG) estimation becomes a challenge when we are not allowed to sample with the target policy but only have access to a dataset generated by some unknown behavior policy. Conventional methods for off-policy PG estimation…

Machine Learning · Statistics 2022-06-22 Chengzhuo Ni , Ruiqi Zhang , Xiang Ji , Xuezhou Zhang , Mengdi Wang

The goal of robust reinforcement learning (RL) is to learn a policy that is robust against the uncertainty in model parameters. Parameter uncertainty commonly occurs in many real-world RL applications due to simulator modeling errors,…

Machine Learning · Computer Science 2022-10-19 Kishan Panaganti , Zaiyan Xu , Dileep Kalathil , Mohammad Ghavamzadeh

We study the off-policy evaluation (OPE) problem in an infinite-horizon Markov decision process with continuous states and actions. We recast the $Q$-function estimation into a special form of the nonparametric instrumental variables (NPIV)…

Statistics Theory · Mathematics 2022-06-28 Xiaohong Chen , Zhengling Qi

Model-free algorithms for reinforcement learning typically require a condition called Bellman completeness in order to successfully operate off-policy with function approximation, unless additional conditions are met. However, Bellman…

Machine Learning · Computer Science 2023-06-07 Andrea Zanette

Offline reinforcement learning learns policies from fixed datasets without further environment interaction. A key challenge in this setting is epistemic uncertainty, arising from limited or biased data coverage, particularly when the…

Machine Learning · Computer Science 2026-04-09 Abhilash Reddy Chenreddy , Erick Delage

Offline model selection (OMS), that is, choosing the best policy from a set of many policies given only logged data, is crucial for applying offline RL in real-world settings. One idea that has been extensively explored is to select…

Machine Learning · Computer Science 2023-10-18 Joshua P. Zitovsky , Daniel de Marchi , Rishabh Agarwal , Michael R. Kosorok

Offline RL algorithms aim to improve upon the behavior policy that produces the collected data while constraining the learned policy to be within the support of the dataset. However, practical offline datasets often contain examples with…

Machine Learning · Computer Science 2026-02-12 Jianxun Wang , Grant C. Forbes , Leonardo Villalobos-Arias , David L. Roberts

It is well known that the extension of Watkins' algorithm to general function approximation settings is challenging: does the projected Bellman equation have a solution? If so, is the solution useful in the sense of generating a good…

Optimization and Control · Mathematics 2020-08-11 Prashant G. Mehta , Sean P. Meyn

Offline reinforcement learning (RL) aims to learn an effective policy from a static dataset. To alleviate extrapolation errors, existing studies often uniformly regularize the value function or policy updates across all states. However, due…

Machine Learning · Computer Science 2025-05-27 Qin-Wen Luo , Ming-Kun Xie , Ye-Wen Wang , Sheng-Jun Huang

Transformer attention is typically implemented using softmax normalization, which enforces attention weights with unit sum normalization. While effective in many settings, this constraint can limit flexibility in controlling attention…

Computation and Language · Computer Science 2026-02-27 Jeongin Bae , Baeseong Park , Gunho Park , Minsub Kim , Joonhyung Lee , Junhee Yoo , Sunghyeon Woo , Jiwon Ryu , Se Jung Kwon , Dongsoo Lee

Q-functions are widely used in discrete-time learning and control to model future costs arising from a given control policy, when the initial state and input are given. Although some of their properties are understood, Q-functions…

Optimization and Control · Mathematics 2019-02-21 Joseph Warrington