arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.03607stat.MLcs.LGmath.STstat.TH
收藏

非策略推理的Bootstrapping拟合Q-估计

Bootstrapping Fitted Q-Evaluation for Off-Policy Inference

Deepmind(深度思维);Princeton University(普林斯顿大学);University of Alberta(阿尔伯塔大学)

Botao Hao, Xiang Ji, Yaqi Duan, Hao Lu, Csaba Szepesvári, Mengdi Wang

摘要:

Bootstrapping提供了一种灵活有效的方法来评估批处理强化学习的质量,但其理论特性还不太清楚。在本文中,我们研究了使用自举在离政策评估(OPE),特别是,我们专注于拟合Q-评估(QFE),这是已知的最小最优的表格和线性模型的情况下。我们提出了一个自举的方法来推断的政策评估误差的分布,并证明了这种方法是渐近有效的和分布一致的政策外统计推断。为了克服自举的计算限制,我们进一步调整子采样过程,提高了一个数量级的运行时间。我们在经典RL环境中数值评估了Boottrapping方法的置信区间估计,估计偏离策略评估器的方差,估计多个偏离策略评估器之间的相关性。

英文摘要:

Bootstrapping provides a flexible and effective approach for assessing the quality of batch reinforcement learning, yet its theoretical property is less understood. In this paper, we study the use of bootstrapping in off-policy evaluation (OPE), and in particular, we focus on the fitted Q-evaluation (FQE) that is known to be minimax-optimal in the tabular and linear-model cases. We propose a bootstrapping FQE method for inferring the distribution of the policy evaluation error and show that this method is asymptotically efficient and distributionally consistent for off-policy statistical inference. To overcome the computation limit of bootstrapping, we further adapt a subsampling procedure that improves the runtime by an order of magnitude. We numerically evaluate the bootrapping method in classical RL environments for confidence interval estimation, estimating the variance of off-policy evaluator, and estimating the correlation between multiple off-policy evaluators.

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。

↑