arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1906.07586cs.LGstat.ML
收藏

有效和噪声容忍强化学习的间隙增加策略评估

Gap-Increasing Policy Evaluation for Efficient and Noise-Tolerant Reinforcement Learning

Okinawa Inst. of Sci. and Tech.(冲绳科学技术大学院大学)

Tadashi Kozuno, Dongqi Han, Kenji Doya

摘要:

在强化学习(RL)的实际应用中,来自环境固有随机性的噪声是不可避免的。然而,目前的策略评估算法,这在许多RL算法中起着关键作用,要么是容易产生噪声或效率低下。为了解决这个问题,我们引入了一种新的策略评估算法,我们称之为间隙增加RetrAce策略评估(GRAPE)。它利用了两个最近的想法:(1)优势学习中的间隙增加值更新算子以实现噪声容忍;(2)Retrace算法中的离线资格跟踪以实现高效学习。我们提供了详细的理论分析的新算法,表明其效率和噪声的宽容继承的回溯和优势学习。此外,我们的分析表明,GRAPE的学习是显着有效的比一个简单的基于学习率的方法,同时保持相同的噪声容忍度。我们将GRAPE应用于控制问题,并获得了支持我们的理论分析的实验结果。

英文摘要:

In real-world applications of reinforcement learning (RL), noise from inherent stochasticity of environments is inevitable. However, current policy evaluation algorithms, which plays a key role in many RL algorithms, are either prone to noise or inefficient. To solve this issue, we introduce a novel policy evaluation algorithm, which we call Gap-increasing RetrAce Policy Evaluation (GRAPE). It leverages two recent ideas: (1) gap-increasing value update operators in advantage learning for noise-tolerance and (2) off-policy eligibility trace in Retrace algorithm for efficient learning. We provide detailed theoretical analysis of the new algorithm that shows its efficiency and noise-tolerance inherited from Retrace and advantage learning. Furthermore, our analysis shows that GRAPE's learning is significantly efficient than that of a simple learning-rate-based approach while keeping the same level of noise-tolerance. We applied GRAPE to control problems and obtained experimental results supporting our theoretical analysis.

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。

↑