{"created":"2025-01-19T01:45:06.177828+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00240742","sets":["6164:6165:6210:11853"]},"path":["11853"],"owner":"44499","recid":"240742","title":["着手確率を用いた人間の勝率予測の改善"],"pubdate":{"attribute_name":"公開日","attribute_value":"2024-11-15"},"_buckets":{"deposit":"defc7e47-d1d6-4ef5-84aa-a564b81f045a"},"_deposit":{"id":"240742","pid":{"type":"depid","value":"240742","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"着手確率を用いた人間の勝率予測の改善","author_link":["660991","660994","660995","660992","660993","660990"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"着手確率を用いた人間の勝率予測の改善"},{"subitem_title":"Improving Accuracy of Win Rate Prediction for Human Players Using Selection Probability of Moves","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"勝率予測","subitem_subject_scheme":"Other"},{"subitem_subject":"ヒューマン AI インタラクション","subitem_subject_scheme":"Other"},{"subitem_subject":"将棋","subitem_subject_scheme":"Other"}]},"item_type_id":"18","publish_date":"2024-11-15","item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_18_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"北陸先端科学技術大学院大学"},{"subitem_text_value":"北陸先端科学技術大学院大学"},{"subitem_text_value":"北陸先端科学技術大学院大学"}]},"item_18_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Japan Advanced Institute of Science and Technology","subitem_text_language":"en"},{"subitem_text_value":"Japan Advanced Institute of Science and Technology","subitem_text_language":"en"},{"subitem_text_value":"Japan Advanced Institute of Science and Technology","subitem_text_language":"en"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/240742/files/IPSJ-GPWS2024024.pdf","label":"IPSJ-GPWS2024024.pdf"},"date":[{"dateType":"Available","dateValue":"2024-11-15"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-GPWS2024024.pdf","filesize":[{"value":"540.3 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"5"},{"tax":["include_tax"],"price":"0","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"18"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"70fdd5e8-7033-4084-a524-43e0e08f1cb8","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2024 by the Information Processing Society of Japan"}]},"item_18_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"小川, 竜欣"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"シュエ, ジュウシュエン"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"池田, 心"}],"nameIdentifiers":[{}]}]},"item_18_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Tatsuyoshi, Ogawa","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Chu-Hsuan, Hsueh","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Kokolo, Ikeda","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_5794","resourcetype":"conference paper"}]},"item_18_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"近年，強いゲーム AI を指導や観戦支援に役立てる研究や， 指導・観戦用のゲーム AI を作る研究が進められている． 指導・観戦の観点から考えると，強いゲーム AI にとっての最善手だけではなく，人間プレイヤがどのような手を指しそうなのか，その場合の人間プレイヤにとっての勝率がいくらなのかを知ることができれば有益である．強いゲーム AI は主に AI 自身の勝利を目標に学習しているため，人間の指し手・ 予測勝率を正確に予測できるとは限らない．人間の指し手を予測する研究は盛んに行われている一方で，人間の勝率を予測する研究には残された課題が多い．本研究では将棋を対象に，既存の AlphaZero のような強化学習手法や教師あり学習手法の人間の勝率予測性能を調べる．そのうえで，着手確率分布を組み合わせることで，人間にとって良さそうに見える手の勝率を高めに，思いつきにくい手の勝率を低めに重みづけて勝率を予測する手法を提案する．実験の結果，提案モデルは既存モデルに比べて，より高精度で勝率を予測できることが分かった．","subitem_description_type":"Other"}]},"item_18_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"In recent years, research has been progressing in utilizing strong game AIs for teaching human players or supporting spectators, as well as in developing game AIs specifically for teaching or spectating purposes. From the perspective of teaching or spectating, it would be beneficial not only to know the optimal moves of strong game AIs but also to understand what moves human players are likely to make and what their win rates would be in such cases. Since strong game AIs are mainly trained with the objective of maximizing their own win rates, they may not necessarily be able to accurately predict human moves or the win rates for human players. While much research has been conducted on predicting human moves, there remain many challenges in predicting human win rates. In this study, we target Shogi and investigate the performance of existing reinforcement learning methods, like AlphaZero, as well as supervised learning methods in predicting human win rates. Furthermore, we propose a method that combines move probability distributions to weight the win rates, assigning higher weights to moves that seem good from human perspectives and lower weights to moves that are harder to come up with. Experimental results show that the proposed model can predict win rates with higher accuracy compared to existing models.","subitem_description_type":"Other"}]},"item_18_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"172","bibliographic_titles":[{"bibliographic_title":"ゲームプログラミングワークショップ2024論文集"}],"bibliographicPageStart":"166","bibliographicIssueDates":{"bibliographicIssueDate":"2024-11-15","bibliographicIssueDateType":"Issued"},"bibliographicVolumeNumber":"2024"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":240742,"updated":"2025-01-19T07:52:50.747817+00:00","links":{}}