arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.02979cs.AI
收藏

Zero-Shot配合的“双打”

"Other-Play" for Zero-Shot Coordination

Facebook AI Research(脸书人工智能研究院)

Hengyuan Hu, Adam Lerer, Alex Peysakhovich, Jakob Foerster

摘要:

我们考虑了zero-shot协调问题-构建能够与他们以前从未见过的新伙伴(例如人类)协调的人工智能智能体。标准的多智能体强化学习(MARL)方法通常关注自我游戏(SP)设置,其中智能体通过重复与自己玩游戏来构建策略。不幸的是,将SP天真地应用于zero-shot协调问题,可能会产生建立高度专业化惯例的代理,这些惯例不会传递给他们没有训练过的新伙伴。我们介绍了一种新的学习算法,称为其他发挥(OP),提高自我发挥寻找更强大的战略,利用存在已知的对称性的基本问题。我们的特点OP理论上以及实验。我们研究了合作纸牌游戏Hanabi,并表明OP代理与独立训练的代理配对时获得更高的分数。在初步结果中,我们还表明,我们的OP代理获得更高的平均分数时,与人类球员配对,相比,国家的最先进的SP代理。

英文摘要:

We consider the problem of zero-shot coordination - constructing AI agents that can coordinate with novel partners they have not seen before (e.g. humans). Standard Multi-Agent Reinforcement Learning (MARL) methods typically focus on the self-play (SP) setting where agents construct strategies by playing the game with themselves repeatedly. Unfortunately, applying SP naively to the zero-shot coordination problem can produce agents that establish highly specialized conventions that do not carry over to novel partners they have not been trained with. We introduce a novel learning algorithm called other-play (OP), that enhances self-play by looking for more robust strategies, exploiting the presence of known symmetries in the underlying problem. We characterize OP theoretically as well as experimentally. We study the cooperative card game Hanabi and show that OP agents achieve higher scores when paired with independently trained agents. In preliminary results we also show that our OP agents obtains higher average scores when paired with human players, compared to state-of-the-art SP agents.

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。

↑