arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.08078cs.LG
收藏

利用双样本检验解决强化学习中的最大化偏差

Addressing Maximization Bias in Reinforcement Learning with Two-Sample Testing

Technische Universität Dresden(德累斯顿工业大学);Center for Scalable Data Analytics and Artificial Intelligence (ScaDS.AI)(可扩展数据分析与人工智能中心(ScaDS.AI))

Martin Waltz, Ostap Okhrin

AI总结:

针对强化学习价值估计中的最大化偏差,提出基于双样本检验的 TE 与推广 KE,并改造 Q-Learning 和 BDQN,实现可调偏差、收敛保证与性能提升。

摘要:

基于价值的强化学习算法已在游戏、机器人和其他现实应用中展现出强劲效果。高估偏差是这些算法的已知威胁,有时会导致性能急剧下降,甚至造成算法完全失效。我们从统计学角度表述该偏差问题,并将其视为估计一组随机变量的最大期望值(MEV)的一个实例。我们提出基于均值双样本检验的 $T$-Estimator(TE),通过调整底层假设检验的显著性水平,在高估与低估之间灵活插值。我们还引入了一种推广形式,称为 $K$-Estimator(KE),它遵循与 TE 相同的偏差和方差界,并依赖几乎任意的核函数。我们介绍了使用 TE 和 KE 对 $Q$-Learning 以及 Bootstrapped Deep $Q$-Network(BDQN)的修改,并证明了其在表格设置下的收敛性。此外,我们提出了基于 TE 的 BDQN 的自适应变体,可动态调整显著性水平以最小化绝对估计偏差。所有提出的估计器和算法都在多样的任务和环境中经过充分测试与验证,展示了 TE 和 KE 在偏差控制与性能方面的潜力。

英文摘要:

Value-based reinforcement-learning algorithms have shown strong results in games, robotics, and other real-world applications. Overestimation bias is a known threat to those algorithms and can sometimes lead to dramatic performance decreases or even complete algorithmic failure. We frame the bias problem statistically and consider it an instance of estimating the maximum expected value (MEV) of a set of random variables. We propose the $T$-Estimator (TE) based on two-sample testing for the mean, that flexibly interpolates between over- and underestimation by adjusting the significance level of the underlying hypothesis tests. We also introduce a generalization, termed $K$-Estimator (KE), that obeys the same bias and variance bounds as the TE and relies on a nearly arbitrary kernel function. We introduce modifications of $Q$-Learning and the Bootstrapped Deep $Q$-Network (BDQN) using the TE and the KE, and prove convergence in the tabular setting. Furthermore, we propose an adaptive variant of the TE-based BDQN that dynamically adjusts the significance level to minimize the absolute estimation bias. All proposed estimators and algorithms are thoroughly tested and validated on diverse tasks and environments, illustrating the bias control and performance potential of the TE and KE.

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。

↑