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 2006.02243cs.LGstat.ML
收藏

价值改进路径:实现强化学习的更好表示

The Value-Improvement Path: Towards Better Representations for Reinforcement Learning

DeepMind(深度思维公司);Google Research(谷歌研究院)

Will Dabney, André Barreto, Mark Rowland, Robert Dadashi, John Quan, Marc G. Bellemare, David Silver

摘要:

在基于值的强化学习(RL)中,与监督学习不同,智能体面临的不是单一的、固定的近似问题,而是一系列的值预测问题。每次政策改善,问题的性质就会改变,国家的分布及其价值观都会发生变化。在本文中,我们采取了一个新的视角,认为RL代理所面临的价值预测问题不应该孤立地解决,而是作为一个单一的,整体的预测问题。RL算法生成一系列策略,这些策略至少近似地朝着最优策略改进。我们显式地描述了相关的价值函数序列,并称之为价值改进路径,我们的主要思想是整体地近似价值改进路径,而不是仅仅跟踪当前策略的价值函数。具体来说,我们讨论了RL的这种整体观点对表示学习的影响。我们证明,跨越过去的价值改进路径的表示也将为未来的政策改进提供准确的价值近似。我们使用这种见解,以更好地了解现有的方法来辅助任务,并提出新的。为了验证我们的假设,我们在一个标准的深度强化学习智能体上增加了一个学习价值提升路径的辅助任务。在一项对Atari 2600游戏的研究中,增强后的智能体的平均和中值性能大约是基线智能体的两倍。

英文摘要:

In value-based reinforcement learning (RL), unlike in supervised learning, the agent faces not a single, stationary, approximation problem, but a sequence of value prediction problems. Each time the policy improves, the nature of the problem changes, shifting both the distribution of states and their values. In this paper we take a novel perspective, arguing that the value prediction problems faced by an RL agent should not be addressed in isolation, but rather as a single, holistic, prediction problem. An RL algorithm generates a sequence of policies that, at least approximately, improve towards the optimal policy. We explicitly characterize the associated sequence of value functions and call it the value-improvement path. Our main idea is to approximate the value-improvement path holistically, rather than to solely track the value function of the current policy. Specifically, we discuss the impact that this holistic view of RL has on representation learning. We demonstrate that a representation that spans the past value-improvement path will also provide an accurate value approximation for future policy improvements. We use this insight to better understand existing approaches to auxiliary tasks and to propose new ones. To test our hypothesis empirically, we augmented a standard deep RL agent with an auxiliary task of learning the value-improvement path. In a study of Atari 2600 games, the augmented agent achieved approximately double the mean and median performance of the baseline agent.

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。

↑