DeltaDou: Expert-Level Doudizhu AI Through Self-Play

Abstract

Artificial Intelligence has seen several breakthroughs in two-player perfect information game.  Nevertheless, Doudizhu, a three-player imperfect information game, is still quite challenging.  In this paper, we present a Doudizhu AI by applying deep reinforcement learning from games of self-play.  The algorithm combines an asymmetric MCTS on nodes of information set of each player, a policy-value network that approximates the policy and value on each decision node, and inference on unobserved hands of other players by given policy.  Our results show that self-play can significantly improve the performance of our agent in this multi-agent imperfect information game.  Even starting with a weak AI, our agent can achieve human expert level after days of self-play and training.

Cite

Text

Jiang et al. "DeltaDou: Expert-Level Doudizhu AI Through Self-Play." International Joint Conference on Artificial Intelligence, 2019. doi:10.24963/IJCAI.2019/176

Markdown

[Jiang et al. "DeltaDou: Expert-Level Doudizhu AI Through Self-Play." International Joint Conference on Artificial Intelligence, 2019.](https://mlanthology.org/ijcai/2019/jiang2019ijcai-deltadou/) doi:10.24963/IJCAI.2019/176

BibTeX

@inproceedings{jiang2019ijcai-deltadou,
  title     = {{DeltaDou: Expert-Level Doudizhu AI Through Self-Play}},
  author    = {Jiang, Qiqi and Li, Kuangzheng and Du, Boyao and Chen, Hao and Fang, Hai},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2019},
  pages     = {1265-1271},
  doi       = {10.24963/IJCAI.2019/176},
  url       = {https://mlanthology.org/ijcai/2019/jiang2019ijcai-deltadou/}
}