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 2102.04881cs.LGcs.CY
收藏

衡量深度强化学习样本效率的进展

Measuring Progress in Deep Reinforcement Learning Sample Efficiency

ETH Zurich(苏黎世联邦理工学院);Institute of Science, Technology and Policy(科学、技术与政策研究所)

Florian E. Dorner

摘要:

采样环境转换是深度强化学习(DRL)算法的关键输入。当前的DRL基准通常允许廉价且容易地生成大量样本,使得DRL中的感知进展不一定对应于改进的样本效率。由于模拟真实世界的过程通常是非常困难的,并且收集真实世界的经验是昂贵的,因此样本效率是DRL的经济相关应用的重要指标。我们调查的进展,样本效率的Atari游戏和连续控制任务,通过比较样本的数量,各种算法需要达到一个给定的性能水平,根据相应的出版物中的训练曲线。我们发现样本效率呈指数级增长,估计Atari的倍增时间约为10至18个月,基于状态的连续控制约为5至24个月,基于像素的连续控制约为4至9个月,具体取决于特定的任务和性能水平。

英文摘要:

Sampled environment transitions are a critical input to deep reinforcement learning (DRL) algorithms. Current DRL benchmarks often allow for the cheap and easy generation of large amounts of samples such that perceived progress in DRL does not necessarily correspond to improved sample efficiency. As simulating real world processes is often prohibitively hard and collecting real world experience is costly, sample efficiency is an important indicator for economically relevant applications of DRL. We investigate progress in sample efficiency on Atari games and continuous control tasks by comparing the number of samples that a variety of algorithms need to reach a given performance level according to training curves in the corresponding publications. We find exponential progress in sample efficiency with estimated doubling times of around 10 to 18 months on Atari, 5 to 24 months on state-based continuous control and of around 4 to 9 months on pixel-based continuous control depending on the specific task and performance level.

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。

↑