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 2107.02729cs.LGcs.AIstat.ML
收藏

AdaRL:迁移强化学习中适应什么、在哪里以及如何适应

AdaRL: What, Where, and How to Adapt in Transfer Reinforcement Learning

Carnegie Mellon University(卡内基梅隆大学);City University of Hong Kong(香港城市大学);University of Cambridge(剑桥大学);Max Planck Institute for Intelligent Systems(马克斯·普朗克智能系统研究所);University of Amsterdam(阿姆斯特丹大学);MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室)

Biwei Huang, Fan Feng, Chaochao Lu, Sara Magliacane, Kun Zhang

摘要:

强化学习(RL)的一个实际挑战是如何在面对新环境时快速适应。在本文中,我们提出了一个原则性的框架,自适应RL,称为\textit{AdaRL},可靠和有效地适应跨域的变化与目标域的几个样本,即使在部分可观察的环境。具体来说,我们利用一个简约的图形表示,其特征在于在RL系统中的变量的结构关系。这种图形表示提供了一种紧凑的方式来编码跨域的变化是什么和在哪里,并进一步告知我们为了策略适应的目的而必须考虑的最小变化集。我们表明,通过明确利用这种紧凑的表示来编码变化,我们可以有效地适应目标域的政策,其中只需要几个样本,并避免进一步的政策优化。我们通过一系列实验说明了AdaRL的功效,这些实验改变了Cartpole和Atari游戏的观察,过渡和奖励功能。

英文摘要:

One practical challenge in reinforcement learning (RL) is how to make quick adaptations when faced with new environments. In this paper, we propose a principled framework for adaptive RL, called \textit{AdaRL}, that adapts reliably and efficiently to changes across domains with a few samples from the target domain, even in partially observable environments. Specifically, we leverage a parsimonious graphical representation that characterizes structural relationships over variables in the RL system. Such graphical representations provide a compact way to encode what and where the changes across domains are, and furthermore inform us with a minimal set of changes that one has to consider for the purpose of policy adaptation. We show that by explicitly leveraging this compact representation to encode changes, we can efficiently adapt the policy to the target domain, in which only a few samples are needed and further policy optimization is avoided. We illustrate the efficacy of AdaRL through a series of experiments that vary factors in the observation, transition, and reward functions for Cartpole and 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。

↑