arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.14604cs.LGcs.AIstat.ML
收藏

数量与质量:深度强化学习的超参数优化

Quantity vs. Quality: On Hyperparameter Optimization for Deep Reinforcement Learning

University of California, Irvine(加州大学尔湾分校)

Lars Hertel, Pierre Baldi, Daniel L. Gillen

摘要:

强化学习算法可以在具有不同随机种子的训练运行之间显示出性能的强烈变化。在本文中,我们将探讨这如何影响超参数优化时,目标是找到超参数设置,以及在随机种子。特别是,我们基准是否是更好地探索大量的超参数设置,通过修剪表现不佳,或者如果它是更好的目标,通过使用重复收集的结果的质量。为此,我们考虑连续减半,随机搜索和贝叶斯优化算法,后两个有和没有重复。我们将这些应用于调整Cartpole平衡任务和倒立摆摆动任务上的PPO2算法。我们证明了修剪可能会对优化产生负面影响,并且重复采样无助于找到在随机种子中表现更好的超参数设置。从我们的实验中,我们得出结论,贝叶斯优化与噪声鲁棒采集功能是强化学习任务中的超参数优化的最佳选择。

英文摘要:

Reinforcement learning algorithms can show strong variation in performance between training runs with different random seeds. In this paper we explore how this affects hyperparameter optimization when the goal is to find hyperparameter settings that perform well across random seeds. In particular, we benchmark whether it is better to explore a large quantity of hyperparameter settings via pruning of bad performers, or if it is better to aim for quality of collected results by using repetitions. For this we consider the Successive Halving, Random Search, and Bayesian Optimization algorithms, the latter two with and without repetitions. We apply these to tuning the PPO2 algorithm on the Cartpole balancing task and the Inverted Pendulum Swing-up task. We demonstrate that pruning may negatively affect the optimization and that repeated sampling does not help in finding hyperparameter settings that perform better across random seeds. From our experiments we conclude that Bayesian optimization with a noise robust acquisition function is the best choice for hyperparameter optimization in reinforcement learning tasks.

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。

↑