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 2102.12425cs.LG
收藏

长期信用分配的综合收益

Synthetic Returns for Long-Term Credit Assignment

DeepMind(深度思考公司)

David Raposo, Sam Ritter, Adam Santoro, Greg Wayne, Theophane Weber, Matt Botvinick, Hado van Hasselt, Francis Song

摘要:

自强化学习的早期以来,随着时间的推移为动作分配信用的主力方法一直是时间差(TD)学习,它逐时间步向后传播信用。当行动和奖励之间的延迟很长,并且当干预的不相关事件对长期回报产生差异时,这种方法就会受到影响。我们提出了状态关联(SA)学习,其中代理学习状态和任意遥远的未来奖励之间的关联,然后直接在两者之间传播信用。在这项工作中,我们使用SA学习来模拟过去状态对当前奖励的贡献。通过这个模型,我们可以预测每个州对未来的贡献,我们称之为“综合回报”。然后,TD学习可以应用于选择最大化这些合成回报(SR)的行动。我们证明了增强代理的有效性与SR在一系列的TD学习单独失败的任务。我们表明,学习SR是可解释的:他们尖峰状态发生后,采取关键行动。最后,我们展示了我们基于IMPALA的SR代理解决Atari Skiing -一个具有漫长奖励延迟的游戏,这对深度RL代理构成了主要障碍-比已发布的最新技术快25倍。

英文摘要:

Since the earliest days of reinforcement learning, the workhorse method for assigning credit to actions over time has been temporal-difference (TD) learning, which propagates credit backward timestep-by-timestep. This approach suffers when delays between actions and rewards are long and when intervening unrelated events contribute variance to long-term returns. We propose state-associative (SA) learning, where the agent learns associations between states and arbitrarily distant future rewards, then propagates credit directly between the two. In this work, we use SA-learning to model the contribution of past states to the current reward. With this model we can predict each state's contribution to the far future, a quantity we call "synthetic returns". TD-learning can then be applied to select actions that maximize these synthetic returns (SRs). We demonstrate the effectiveness of augmenting agents with SRs across a range of tasks on which TD-learning alone fails. We show that the learned SRs are interpretable: they spike for states that occur after critical actions are taken. Finally, we show that our IMPALA-based SR agent solves Atari Skiing -- a game with a lengthy reward delay that posed a major hurdle to deep-RL agents -- 25 times faster than the published state-of-the-art.

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。

↑