arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.09533cs.AI
收藏

在StarCraft多智能体挑战赛中,您只需要独立学习吗?

Is Independent Learning All You Need in the StarCraft Multi-Agent Challenge?

University of Oxford(牛津大学);Snap Inc.(森宝公司);NVIDIA(英伟达公司)

Christian Schroeder de Witt, Tarun Gupta, Denys Makoviichuk, Viktor Makoviychuk, Philip H. S. Torr, Mingfei Sun, Shimon Whiteson

摘要:

在\emph{centralized training with decentralized execution}环境中,最近开发的合作多智能体强化学习方法涉及估计集中式联合价值函数。在本文中,我们证明,尽管其各种理论上的缺点,独立PPO(IPPO),独立学习的形式,其中每个代理简单地估计其本地价值函数,可以执行一样好或更好地比国家的最先进的联合学习方法流行的多代理基准套件SMAC的超参数调整。我们还比较了IPPO的几个变种,结果表明,IPPO的强大性能可能是由于它的鲁棒性,某些形式的环境非平稳性。

英文摘要:

Most recently developed approaches to cooperative multi-agent reinforcement learning in the \emph{centralized training with decentralized execution} setting involve estimating a centralized, joint value function. In this paper, we demonstrate that, despite its various theoretical shortcomings, Independent PPO (IPPO), a form of independent learning in which each agent simply estimates its local value function, can perform just as well as or better than state-of-the-art joint learning approaches on popular multi-agent benchmark suite SMAC with little hyperparameter tuning. We also compare IPPO to several variants; the results suggest that IPPO's strong performance may be due to its robustness to some forms of environment non-stationarity.

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。

↑