Werewolf game modeling using action probabilities based on play log analysis

Yuya Hirata, Michimasa Inaba, Kenichi Takahashi, Fujio Toriumi, Hirotaka Osawa, Daisuke Katagami, Kousuke Shinoda

研究成果: Conference contribution

8 被引用数 (Scopus)

抄録

In this study, we construct a non-human agent that can play the werewolf game (i.e., AI wolf) with aims of creating more advanced intelligence and acquire more advanced communication skills for AIbased systems. We therefore constructed a behavioral model using information regarding human players and the decisions made by such players; all such information was obtained from play logs of the werewolf game. To confirm our model, we conducted simulation experiments of the werewolf game using an agent based on our proposed behavioral model, as well as a random agent for comparison. Consequently, we obtained an 81.55% coincidence ratio of agent behavior versus human behavior.

本文言語English
ホスト出版物のタイトルComputers and Games - 9th International Conference, CG 2016, Revised Selected Papers
編集者Aske Plaat, Walter Kosters, Jaap van den Herik
出版社Springer Verlag
ページ103-114
ページ数12
ISBN(印刷版)9783319509341
DOI
出版ステータスPublished - 2016
外部発表はい
イベント9th International Conference on Computer and Games, CG 2016 - Leiden, Netherlands
継続期間: 2016 6月 292016 7月 1

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
10068 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Conference

Conference9th International Conference on Computer and Games, CG 2016
国/地域Netherlands
CityLeiden
Period16/6/2916/7/1

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)

フィンガープリント

「Werewolf game modeling using action probabilities based on play log analysis」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル