arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2002.06487cs.LGcs.AI
收藏

最大最小Q学习:控制Q学习的估计偏差

Maxmin Q-learning: Controlling the Estimation Bias of Q-learning

University of Alberta(阿尔伯塔大学)

Qingfeng Lan, Yangchen Pan, Alona Fyshe, Martha White

摘要:

Q学习存在高估偏差,因为它使用最大估计动作值来近似最大动作值。已经提出了一些算法来减少高估偏差,但我们缺乏对偏差如何与性能相互作用的理解,以及现有算法在多大程度上减轻偏差。在本文中,我们1)强调高估偏差对学习效率的影响是环境依赖的; 2)提出了一种推广的Q-学习,称为\emph{Maxmin Q-learning},它提供了一个参数来灵活地控制偏差; 3)从理论上证明了存在一个参数选择的最大最小Q-学习,导致无偏估计比Q-学习更低的近似方差; 4)利用一种新的广义Q学习框架,证明了我们的算法在表格情形下的收敛性,以及几种已有的Q学习算法的收敛性。我们的经验验证,我们的算法更好地控制估计偏差在玩具环境中,它实现了几个基准问题的优异性能。

英文摘要:

Q-learning suffers from overestimation bias, because it approximates the maximum action value using the maximum estimated action value. Algorithms have been proposed to reduce overestimation bias, but we lack an understanding of how bias interacts with performance, and the extent to which existing algorithms mitigate bias. In this paper, we 1) highlight that the effect of overestimation bias on learning efficiency is environment-dependent; 2) propose a generalization of Q-learning, called \emph{Maxmin Q-learning}, which provides a parameter to flexibly control bias; 3) show theoretically that there exists a parameter choice for Maxmin Q-learning that leads to unbiased estimation with a lower approximation variance than Q-learning; and 4) prove the convergence of our algorithm in the tabular case, as well as convergence of several previous Q-learning variants, using a novel Generalized Q-learning framework. We empirically verify that our algorithm better controls estimation bias in toy environments, and that it achieves superior performance on several benchmark problems.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑