{"created":"2023-05-15T08:43:47.329698+00:00","id":8875,"links":{},"metadata":{"_buckets":{"deposit":"c4f53570-873f-449e-a00a-595f77d7a3af"},"_deposit":{"created_by":13,"id":"8875","owners":[13],"pid":{"revision_id":0,"type":"depid","value":"8875"},"status":"published"},"_oai":{"id":"oai:uec.repo.nii.ac.jp:00008875","sets":["6"]},"author_link":["24137","24138","24139"],"control_number":"8875","item_10001_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"2018-05-01","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"5","bibliographicPageEnd":"768","bibliographicPageStart":"754","bibliographicVolumeNumber":"J101-D","bibliographic_titles":[{"bibliographic_title":"電子情報通信学会論文誌. D, 情報・システム","bibliographic_titleLang":"ja"}]}]},"item_10001_description_5":{"attribute_name":"抄録","attribute_value_mlt":[{"subitem_description":"漸近一致性をもつベイジアンネットワークの構造学習はNP困難である．これまで動的計画法やA*探索，整数計画法による探索アルゴリズムが開発されてきたが，未だに60ノード程度の構造学習を限界とし，大規模構造学習の実現のためには，全く異なるアプローチの開発が急務である．一方で因果モデルの研究分野では，条件付き独立性テスト（CIテスト）と方向付けによる画期的に計算量を削減した構造学習アプローチが提案されている．このアプローチは制約ベースアプローチと呼ばれ，RAIアルゴリズムが最も高精度な最先端学習法として知られている．しかしRAIアルゴリズムは，CIテストに仮説検定法または条件付き相互情報量を用いている．前者の精度は帰無仮説が正しい確率を表すp値とユーザが設定する有意水準に依存する．p値はデータ数の増加により小さい値を取り，誤って帰無仮説を棄却してしまう問題が知られている．一方で，後者の精度はしきい値の設定に強く影響する．したがって，漸近的に真の構造を学習できる保証がない．本論文では，漸近一致性を有するBayes factorを用いたCIテストをRAIアルゴリズムに組み込む．これにより，数百ノードをもつ大規模構造学習を実現する．数種類のベンチマークネットワークを用いたシミュレーション実験により，本手法の有意性を示す．","subitem_description_type":"Abstract"},{"subitem_description":"A score-based learning Bayesian networks is NP-hard. On the other hands, constraint-based approach, that can dynamically relaxes the computational cost, is applicable to learning huge Bayesian network structures. The approach uses conditional independence (CI) tests based on the conditional mutual information and statistical testings. However, those CI tests have no consistency. In this paper, we propose a new constraint-based learning method that uses the CI test based on the Bayes factor, which have consistency. The proposed method combines it to the RAI algorithm, that is a state-of-the-art algorithm of the constraint-based approach. The experimental result shows our proposed method provides empirically best performance.","subitem_description_type":"Abstract"}]},"item_10001_publisher_8":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"電子情報通信学会"}]},"item_10001_relation_14":{"attribute_name":"DOI","attribute_value_mlt":[{"subitem_relation_type":"isIdenticalTo","subitem_relation_type_id":{"subitem_relation_type_id_text":"10.14923/transinfj.2017JDP7089","subitem_relation_type_select":"DOI"}}]},"item_10001_relation_17":{"attribute_name":"関連サイト","attribute_value_mlt":[{"subitem_relation_type_id":{"subitem_relation_type_id_text":"http://search.ieice.org/index.html","subitem_relation_type_select":"URI"}}]},"item_10001_rights_15":{"attribute_name":"権利","attribute_value_mlt":[{"subitem_rights":"Copyright © 2018 IEICE"}]},"item_10001_source_id_9":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"1881-0225","subitem_source_identifier_type":"ISSN"}]},"item_10001_version_type_20":{"attribute_name":"著者版フラグ","attribute_value_mlt":[{"subitem_version_resource":"http://purl.org/coar/version/c_970fb48d4fbd8a85","subitem_version_type":"VoR"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"名取, 和樹","creatorNameLang":"ja"},{"creatorName":"ナトリ, カズキ","creatorNameLang":"ja-Kana"},{"creatorName":"NATORI, Kazuki","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"宇都, 雅輝","creatorNameLang":"ja"},{"creatorName":"ウト, マサキ","creatorNameLang":"ja-Kana"},{"creatorName":"UTO, Masaki","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"植野, 真臣","creatorNameLang":"ja"},{"creatorName":"ウエノ, マオミ","creatorNameLang":"ja-Kana"},{"creatorName":"UENO, Maomi","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2019-01-15"}],"displaytype":"detail","filename":"j101-d_5_754.pdf","filesize":[{"value":"761.5 kB"}],"format":"application/pdf","licensetype":"license_note","mimetype":"application/pdf","url":{"label":"j101-d_5_754","url":"https://uec.repo.nii.ac.jp/record/8875/files/j101-d_5_754.pdf"},"version_id":"fd088ea3-0865-4c8f-a907-b063347cdc29"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"ベイジアンネットワーク","subitem_subject_scheme":"Other"},{"subitem_subject":"確率的グラフィカルモデル","subitem_subject_scheme":"Other"},{"subitem_subject":"構造学習","subitem_subject_scheme":"Other"},{"subitem_subject":"条件付き独立性検定","subitem_subject_scheme":"Other"},{"subitem_subject":"learning Bayesian networks","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"probabilistic graphical models","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"conditional independence tests","subitem_subject_language":"en","subitem_subject_scheme":"Other"},{"subitem_subject":"statistical testings","subitem_subject_language":"en","subitem_subject_scheme":"Other"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"journal article","resourceuri":"http://purl.org/coar/resource_type/c_6501"}]},"item_title":"Bayes factorを用いたRAIアルゴリズムによる大規模ベイジアンネットワーク学習","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"Bayes factorを用いたRAIアルゴリズムによる大規模ベイジアンネットワーク学習","subitem_title_language":"ja"},{"subitem_title":"Learning Huge Bayesian Networks by RAI Algorithm Using Bayes Factor","subitem_title_language":"en"}]},"item_type_id":"10001","owner":"13","path":["6"],"pubdate":{"attribute_name":"PubDate","attribute_value":"2019-01-15"},"publish_date":"2019-01-15","publish_status":"0","recid":"8875","relation_version_is_last":true,"title":["Bayes factorを用いたRAIアルゴリズムによる大規模ベイジアンネットワーク学習"],"weko_creator_id":"13","weko_shared_id":-1},"updated":"2024-03-05T06:20:46.528459+00:00"}