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 1509.03005cs.LGcs.AIcs.NEstat.ML
收藏

连续深度策略强化学习的相容值矩阵

Compatible Value Gradients for Reinforcement Learning of Continuous Deep Policies

Victoria University of Wellington(惠灵顿维多利亚大学)

David Balduzzi, Muhammad Ghifary

摘要:

本文提出了GProp,一个深度强化学习算法的连续政策与兼容函数近似。该算法是基于两个创新。首先,我们提出了一种基于时间差的学习值函数梯度的方法。其次,我们提出了偏离者-行动者-批评者(DAC)模型,它包括三个神经网络,估计的价值函数,其梯度,并确定行动者的政策分别。我们在两个具有挑战性的任务上评估GProp:一个由非参数回归数据集构建的上下文强盗问题,该问题旨在探索强化学习算法准确估计梯度的能力;以及章鱼臂,一个具有挑战性的强化学习基准。GProp在强盗任务上与完全监督方法竞争,并在章鱼臂上实现了迄今为止的最佳性能。

英文摘要:

This paper proposes GProp, a deep reinforcement learning algorithm for continuous policies with compatible function approximation. The algorithm is based on two innovations. Firstly, we present a temporal-difference based method for learning the gradient of the value-function. Secondly, we present the deviator-actor-critic (DAC) model, which comprises three neural networks that estimate the value function, its gradient, and determine the actor's policy respectively. We evaluate GProp on two challenging tasks: a contextual bandit problem constructed from nonparametric regression datasets that is designed to probe the ability of reinforcement learning algorithms to accurately estimate gradients; and the octopus arm, a challenging reinforcement learning benchmark. GProp is competitive with fully supervised methods on the bandit task and achieves the best performance to date on the octopus arm.

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。

↑