arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1502.05477cs.LG
收藏

信任域策略优化

Trust Region Policy Optimization

University of California, Berkeley(加州大学伯克利分校)

John Schulman, Sergey Levine, Philipp Moritz, Michael I. Jordan, Pieter Abbeel

摘要:

我们描述了一个迭代过程优化政策,保证单调改善。通过对理论上合理的过程进行几次近似,我们开发了一种实用的算法,称为信任域策略优化(TRPO)。该算法类似于自然策略梯度方法,对优化神经网络等大型非线性策略是有效的。我们的实验证明了它在各种任务上的强大性能:学习模拟机器人游泳,跳跃和行走步态;以及使用屏幕图像作为输入玩Atari游戏。尽管它的近似偏离理论,TRPO倾向于给出单调的改进,几乎没有调整超参数。

英文摘要:

We describe an iterative procedure for optimizing policies, with guaranteed monotonic improvement. By making several approximations to the theoretically-justified procedure, we develop a practical algorithm, called Trust Region Policy Optimization (TRPO). This algorithm is similar to natural policy gradient methods and is effective for optimizing large nonlinear policies such as neural networks. Our experiments demonstrate its robust performance on a wide variety of tasks: learning simulated robotic swimming, hopping, and walking gaits; and playing Atari games using images of the screen as input. Despite its approximations that deviate from the theory, TRPO tends to give monotonic improvement, with little tuning of hyperparameters.

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。

↑