arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1802.01561cs.LGcs.AI
收藏

IMPALA:具有重要性加权的Actor-Learner架构的可扩展分布式深度RL

IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures

DeepMind Technologies(DeepMind科技公司)

Lasse Espeholt, Hubert Soyer, Remi Munos, Karen Simonyan, Volodymir Mnih, Tom Ward, Yotam Doron, Vlad Firoiu, Tim Harley, Iain Dunning, Shane Legg, Koray Kavukcuoglu

摘要:

在这项工作中,我们的目标是解决一个大的任务集合使用一个单一的强化学习代理与一组参数。一个关键的挑战是处理增加的数据量和延长的训练时间。我们已经开发了一种新的分布式代理IMPALA(重要性加权的演员-学习者架构),它不仅在单机训练中更有效地使用资源,而且可以扩展到数千台机器,而不会牺牲数据效率或资源利用率。我们实现了稳定的学习,在高吞吐量相结合的解耦代理和学习与一种新的偏离政策的校正方法,称为V-跟踪。我们在DMLab-30(来自DeepMind实验室环境的一组30个任务)上证明了IMPAL用于多任务强化学习的有效性(Beattie等人,2016))和Atari-57(Arcade Learning Environment中所有可用的Atari游戏(Bellemare等人,2013年a))。我们的研究结果表明,IMPALA是能够实现更好的性能比以前的代理与更少的数据,关键是表现出积极的任务之间的转移,由于其多任务的方法。

英文摘要:

In this work we aim to solve a large collection of tasks using a single reinforcement learning agent with a single set of parameters. A key challenge is to handle the increased amount of data and extended training time. We have developed a new distributed agent IMPALA (Importance Weighted Actor-Learner Architecture) that not only uses resources more efficiently in single-machine training but also scales to thousands of machines without sacrificing data efficiency or resource utilisation. We achieve stable learning at high throughput by combining decoupled acting and learning with a novel off-policy correction method called V-trace. We demonstrate the effectiveness of IMPALA for multi-task reinforcement learning on DMLab-30 (a set of 30 tasks from the DeepMind Lab environment (Beattie et al., 2016)) and Atari-57 (all available Atari games in Arcade Learning Environment (Bellemare et al., 2013a)). Our results show that IMPALA is able to achieve better performance than previous agents with less data, and crucially exhibits positive transfer between tasks as a result of its multi-task approach.

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。

↑