arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.00445cs.LGcs.AIstat.ML
收藏

Q-学习的Ensemble Bootstrapping

Ensemble Bootstrapping for Q-Learning

Viterbi Faculty of Electrical Engineering, Technion Institute of Technology(以色列理工学院电气工程维特比学院)

Oren Peer, Chen Tessler, Nadav Merlis, Ron Meir

摘要:

Q学习(QL)是一种常用的强化学习算法,由于最优Bellman算子中的最大化项而存在过估计偏差。这种偏差可能导致次优行为。双Q学习通过使用两个估计器来解决这个问题,但会导致低估偏差。类似于Q学习中的过度估计,在某些情况下,估计不足的偏差可能会降低性能。在这项工作中,我们引入了一种新的偏差减少算法,称为Ensemble Bootstrapped Q-Learning(EBQL),这是Double-Q-learning对集成的自然扩展。我们分析我们的方法从理论和经验。理论上,我们证明了EBQL类更新产生较低的MSE时,估计一组独立的随机变量的最大均值。从经验上讲,我们表明,存在域的估计和估计不足的结果在次优性能。最后,我们展示了一套ATARI游戏的EBQL的深度RL变体比其他深度QL算法的优越性能。

英文摘要:

Q-learning (QL), a common reinforcement learning algorithm, suffers from over-estimation bias due to the maximization term in the optimal Bellman operator. This bias may lead to sub-optimal behavior. Double-Q-learning tackles this issue by utilizing two estimators, yet results in an under-estimation bias. Similar to over-estimation in Q-learning, in certain scenarios, the under-estimation bias may degrade performance. In this work, we introduce a new bias-reduced algorithm called Ensemble Bootstrapped Q-Learning (EBQL), a natural extension of Double-Q-learning to ensembles. We analyze our method both theoretically and empirically. Theoretically, we prove that EBQL-like updates yield lower MSE when estimating the maximal mean of a set of independent random variables. Empirically, we show that there exist domains where both over and under-estimation result in sub-optimal performance. Finally, We demonstrate the superior performance of a deep RL variant of EBQL over other deep QL algorithms for a suite of ATARI games.

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。

↑