arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1808.03030cs.LGstat.ML
收藏

基于Wasserstein梯度流的策略优化

Policy Optimization as Wasserstein Gradient Flows

Duke University(杜克大学);SUNY at Buffalo(纽约州立大学布法罗分校)

Ruiyi Zhang, Changyou Chen, Chunyuan Li, Lawrence Carin

摘要:

策略优化是强化学习(RL)的核心组成部分,并且大多数现有的RL方法基于最大化期望总回报或其代理来直接优化策略的参数。虽然经常取得令人鼓舞的经验成功,其基本的数学原理{ \em policy-distribution}优化是不清楚的。我们把政策优化的概率措施的空间,并解释为Wasserstein梯度流。在概率测度空间上,在一定条件下,策略优化问题就成为一个关于分配优化的凸问题。为了使优化可行,我们开发了有效的算法,通过数值求解相应的离散梯度流。我们的技术适用于几个RL设置,并与许多国家的最先进的政策优化算法。实验结果验证了我们的框架的有效性,往往获得更好的性能相比,相关的算法。

英文摘要:

Policy optimization is a core component of reinforcement learning (RL), and most existing RL methods directly optimize parameters of a policy based on maximizing the expected total reward, or its surrogate. Though often achieving encouraging empirical success, its underlying mathematical principle on {\em policy-distribution} optimization is unclear. We place policy optimization into the space of probability measures, and interpret it as Wasserstein gradient flows. On the probability-measure space, under specified circumstances, policy optimization becomes a convex problem in terms of distribution optimization. To make optimization feasible, we develop efficient algorithms by numerically solving the corresponding discrete gradient flows. Our technique is applicable to several RL settings, and is related to many state-of-the-art policy-optimization algorithms. Empirical results verify the effectiveness of our framework, often obtaining better performance compared to related algorithms.

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。

↑