Presentation 2022-09-16
Temporal Modeling of Players for Multi-agent Coordination in Non-Cooperative Game
Junjie Zhong, Toshiharu Sugawara,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Multi-agent interaction structures that contain mixed cooperative-competitive relationships appear in many realistic situations. However, in multi-agent reinforcement learning, how agents recognize other agents and identify coordinated actions with limited information exchange is a challenging issue. We think that when temporal information including other agents' histories (such as their decisions and solutions) and distances from them is shared among agents through communications, all agents can have models of all other agents and thus obtain information necessary for determining coordinated actions through an asymmetric game process. Based on this consideration, our proposed method can mitigate the issue of non-stationarity and describe relationships between agents by setting non-cooperative objectives for the models of other agents to obtain better solutions. We then experimentally show that agents with our method could achieve better coordination and achieve more appropriate behavior than those with the method without communication and the method without modeling other agents.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) game theory / multi-agent reinforcement learning / modeling other agents
Paper # AI2022-26
Date of Issue 2022-09-08 (AI)

Conference Information
Committee AI
Conference Date 2022/9/15(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Yuichi Sei(Univ. of Electro-Comm.)
Vice Chair Yuko Sakurai(AIST) / Tadachika Ozono(Nagoya Inst. of Tech.)
Secretary Yuko Sakurai(Tokyo Univ. of Agriculture and Technology) / Tadachika Ozono(Toho Univ.)
Assistant Kazutaka Matsuzaki(Chuo Univ.)

Paper Information
Registration To Technical Committee on Artificial Intelligence and Knowledge-Based Processing
Language ENG-JTITLE
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Temporal Modeling of Players for Multi-agent Coordination in Non-Cooperative Game
Sub Title (in English)
Keyword(1) game theory
Keyword(2) multi-agent reinforcement learning
Keyword(3) modeling other agents
1st Author's Name Junjie Zhong
1st Author's Affiliation Waseda University(Waseda Univ.)
2nd Author's Name Toshiharu Sugawara
2nd Author's Affiliation Waseda University(Waseda Univ.)
Date 2022-09-16
Paper # AI2022-26
Volume (vol) vol.122
Number (no) AI-186
Page pp.pp.48-53(AI),
#Pages 6
Date of Issue 2022-09-08 (AI)