arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.11513cs.LG
收藏

混合策略梯度:由数据与模型联合驱动的离线策略强化学习

Mixed Policy Gradient: off-policy reinforcement learning driven jointly by data and model

School of Vehicle and Mobility, Tsinghua University(清华大学车辆与出行学院);Institute for Interdisciplinary Information Sciences, Tsinghua University(清华大学交叉信息研究院)

Yang Guan, Jingliang Duan, Shengbo Eben Li, Jie Li, Jianyu Chen, Bo Cheng

AI总结:

本文提出混合策略梯度(MPG)算法,通过加权融合数据驱动与模型驱动的策略梯度,并基于误差上界设计权重调整规则,在不损失渐近性能的前提下加速强化学习收敛。

摘要:

强化学习(RL)在序列决策中展现出巨大潜力。目前,主流RL算法是数据驱动的,与模型驱动方法相比,它们通常具有更好的渐近性能,但收敛速度要慢得多。本文提出了混合策略梯度(MPG)算法,该算法在策略梯度(PG)中融合经验数据与转移模型,以在不降低性能的情况下加速收敛。形式上,MPG被构造为数据驱动PG与模型驱动PG的加权平均,其中前者是学习到的Q值函数的导数,后者是模型预测回报的导数。为了指导权重设计,我们分析并比较了每个PG误差的上界。基于此,采用一种基于规则的方法来启发式地调整权重。特别地,为了获得更好的PG,数据驱动PG的权重被设计为随着学习过程增长,而另一个权重则减小。仿真结果表明,与其他基线算法相比,MPG方法实现了最佳的渐近性能和收敛速度。

英文摘要:

Reinforcement learning (RL) shows great potential in sequential decision-making. At present, mainstream RL algorithms are data-driven, which usually yield better asymptotic performance but much slower convergence compared with model-driven methods. This paper proposes mixed policy gradient (MPG) algorithm, which fuses the empirical data and the transition model in policy gradient (PG) to accelerate convergence without performance degradation. Formally, MPG is constructed as a weighted average of the data-driven and model-driven PGs, where the former is the derivative of the learned Q-value function, and the latter is that of the model-predictive return. To guide the weight design, we analyze and compare the upper bound of each PG error. Relying on that, a rule-based method is employed to heuristically adjust the weights. In particular, to get a better PG, the weight of the data-driven PG is designed to grow along the learning process while the other to decrease. Simulation results show that the MPG method achieves the best asymptotic performance and convergence speed compared with other baseline algorithms.

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。

↑