arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.03236cs.LGcs.ROstat.ML
收藏

矩与匹配:缩小模仿差距的博弈论框架

Of Moments and Matching: A Game-Theoretic Framework for Closing the Imitation Gap

Carnegie Mellon University(卡内基梅隆大学);Aurora Innovation(奥罗拉创新公司)

Gokul Swamy, Sanjiban Choudhury, J. Andrew Bagnell, Zhiwei Steven Wu

摘要:

我们提供了一个统一的观点,以前的模仿学习算法通过镜头的时刻匹配的大家庭。在其核心,我们的分类方案是基于学习者是否试图匹配(1)奖励或(2)专家行为的行动价值时刻,每个选项导致不同的算法方法。通过考虑学习者和专家行为之间的对立选择分歧,我们能够推导出适用于这些类中的所有算法的政策性能的界限,这是我们所知的第一个。我们还介绍了瞬间可恢复性的概念,隐含在许多以前的分析模仿学习,这使我们能够清楚地描绘如何以及每个算法家族能够减轻复合错误。我们推导出三种新的算法模板(AdVIL,AdRIL和DAeQuIL)具有强有力的保证,简单的实现,和竞争力的经验性能。

英文摘要:

We provide a unifying view of a large family of previous imitation learning algorithms through the lens of moment matching. At its core, our classification scheme is based on whether the learner attempts to match (1) reward or (2) action-value moments of the expert's behavior, with each option leading to differing algorithmic approaches. By considering adversarially chosen divergences between learner and expert behavior, we are able to derive bounds on policy performance that apply for all algorithms in each of these classes, the first to our knowledge. We also introduce the notion of moment recoverability, implicit in many previous analyses of imitation learning, which allows us to cleanly delineate how well each algorithmic family is able to mitigate compounding errors. We derive three novel algorithm templates (AdVIL, AdRIL, and DAeQuIL) with strong guarantees, simple implementation, and competitive empirical performance.

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。

↑