Show simple item record

dc.contributor.authorZhao, Xiangyu
dc.contributor.authorHolden, Sean
dc.date.accessioned2022-07-18T23:30:14Z
dc.date.available2022-07-18T23:30:14Z
dc.identifier.urihttps://www.repository.cam.ac.uk/handle/1810/339217
dc.description.abstractMahjong is a multi-player imperfect-information game with challenging features for AI research. Sanma, being a 3-player variant of Japanese Riichi Mahjong, possesses unique characteristics and a more aggressive playing style than the 4- player game. It is thus challenging and of research interest in its own right, but has not been explored. We present Meowjong, the first ever AI for Sanma using deep reinforcement learning (RL). We define a 2-dimensional data structure for encoding the observable information in a game. We pre-train 5 convolutional neural networks (CNNs) for Sanma’s 5 actions—discard, Pon, Kan, Kita and Riichi, and enhance the major (discard) action’s model via self-play reinforcement learning. Meowjong demon- strates potential for becoming the state-of-the-art in Sanma, by achieving test accuracies comparable with AIs for 4-player Mahjong through supervised learning, and gaining a significant further enhancement from reinforcement learning.
dc.rightsAttribution 4.0 International
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.titleTowards a Competitive 3-Player Mahjong AI using Deep Reinforcement Learning
dc.typeConference Object
dc.publisher.departmentDepartment of Computer Science And Technology
dc.date.updated2022-06-27T14:28:11Z
dc.identifier.doi10.17863/CAM.86627
dcterms.dateAccepted2022-06-14
rioxxterms.versionofrecord10.17863/CAM.86627
rioxxterms.versionAM
dc.contributor.orcidHolden, Sean [0000-0001-7979-1148]
pubs.conference-name2022 IEEE Conference on Games
pubs.conference-start-date2022-08-21
cam.depositDate2022-06-27
pubs.conference-finish-date2022-08-24
pubs.licence-identifierapollo-deposit-licence-2-1
pubs.licence-display-nameApollo Repository Deposit Licence Agreement


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record

Attribution 4.0 International
Except where otherwise noted, this item's licence is described as Attribution 4.0 International