arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2201.12518cs.LGcs.AIcs.SYeess.SY
收藏

零阶演员-评论家:一种面向序贯决策问题的演化框架

Zeroth-Order Actor-Critic: An Evolutionary Framework for Sequential Decision Problems

The University of Hong Kong(香港大学);Tsinghua University(清华大学);SunRising AI Ltd.(森瑞昇人工智能有限公司);Institute for Interdisciplinary Information Sciences, Tsinghua University(清华大学交叉信息研究院);Shanghai Qi Zhi Institute(上海期智研究院)

Yuheng Lei, Yao Lyu, Guojian Zhan, Tao Zhang, Jiangtao Li, Jianyu Chen, Shengbo Eben Li, Sifa Zheng

AI总结:

针对演化算法样本复杂度高、强化学习仅适用于可微策略且易陷局部最优的问题,提出零阶演员-评论家演化框架,结合参数空间逐步探索与演员-评论家架构,在多类序贯决策任务上取得优异性能。

摘要:

演化算法(EAs)通过将序贯决策问题(SDPs)简化为静态优化问题,并以零阶方式搜索最优策略参数,已在求解序贯决策问题方面展现出潜力。尽管这类方法通用性极强,但由于忽略了底层的时序结构,往往存在样本复杂度高的问题。相比之下,强化学习(RL)方法通常将序贯决策问题建模为马尔可夫决策过程(MDP)。虽然强化学习方法的样本效率高于演化算法,但它们仅适用于可微策略,且容易陷入局部最优。为解决这些问题,我们提出了一种全新的演化框架——零阶演员-评论家(Zeroth-Order Actor-Critic,ZOAC)。我们提出在参数空间中采用逐步探索的方式,并从理论上推导了零阶策略梯度。我们进一步利用演员-评论家架构,有效利用序贯决策问题的马尔可夫性质,降低梯度估计器的方差。在每次迭代中,ZOAC采用采样器通过参数空间探索收集轨迹,并在一阶策略评估(PEV)和零阶策略改进(PIM)之间交替进行。为评估ZOAC的有效性,我们将其应用于一项具有挑战性的多车道驾驶任务,对一种基于规则的不可微驾驶策略的参数进行优化,该策略包含行为选择、路径规划和轨迹跟踪三个子模块。我们还在三个Gymnasium任务上将其与基于梯度的强化学习方法进行对比,优化具有数千个参数的神经网络策略。实验结果证明了ZOAC在求解序贯决策问题方面的强大能力。在所有任务的总平均回报指标上,ZOAC显著优于将问题视为静态优化的演化算法,并且即使没有一阶信息,其性能也能与基于梯度的强化学习方法相当。

英文摘要:

Evolutionary algorithms (EAs) have shown promise in solving sequential decision problems (SDPs) by simplifying them to static optimization problems and searching for the optimal policy parameters in a zeroth-order way. While these methods are highly versatile, they often suffer from high sample complexity due to their ignorance of the underlying temporal structures. In contrast, reinforcement learning (RL) methods typically formulate SDPs as Markov Decision Process (MDP). Although more sample efficient than EAs, RL methods are restricted to differentiable policies and prone to getting stuck in local optima. To address these issues, we propose a novel evolutionary framework Zeroth-Order Actor-Critic (ZOAC). We propose to use step-wise exploration in parameter space and theoretically derive the zeroth-order policy gradient. We further utilize the actor-critic architecture to effectively leverage the Markov property of SDPs and reduce the variance of gradient estimators. In each iteration, ZOAC employs samplers to collect trajectories with parameter space exploration, and alternates between first-order policy evaluation (PEV) and zeroth-order policy improvement (PIM). To evaluate the effectiveness of ZOAC, we apply it to a challenging multi-lane driving task, optimizing the parameters in a rule-based, non-differentiable driving policy that consists of three sub-modules: behavior selection, path planning, and trajectory tracking. We also compare it with gradient-based RL methods on three Gymnasium tasks, optimizing neural network policies with thousands of parameters. Experimental results demonstrate the strong capability of ZOAC in solving SDPs. ZOAC significantly outperforms EAs that treat the problem as static optimization and matches the performance of gradient-based RL methods even without first-order information, in terms of total average return across all 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。

↑