{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00184929","sets":["934:1022:9058:9280"]},"path":["9280"],"owner":"11","recid":"184929","title":["制限付き識別ランダムウォークによるグラフベースのラベル拡張"],"pubdate":{"attribute_name":"公開日","attribute_value":"2017-12-13"},"_buckets":{"deposit":"763831ca-0bd7-4ef6-8c8e-713b933827e0"},"_deposit":{"id":"184929","pid":{"type":"depid","value":"184929","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"制限付き識別ランダムウォークによるグラフベースのラベル拡張","author_link":["409767","409770","409768","409769"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"制限付き識別ランダムウォークによるグラフベースのラベル拡張"},{"subitem_title":"Graph-based Label Extension Using Restricted Discriminative-random Walk","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"[テクニカルノート] グラフ構造，ランダムウォーク，ラベル拡張，半教師あり学習","subitem_subject_scheme":"Other"}]},"item_type_id":"3","publish_date":"2017-12-13","item_3_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"筑波大学情報学群知識情報・図書館学類"},{"subitem_text_value":"筑波大学図書館情報メディア系"}]},"item_3_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"College of Knowledge and Library Sciences, School of Informatics, University of Tsukuba","subitem_text_language":"en"},{"subitem_text_value":"Faculty of Library, Information and Media Science, University of Tsukuba","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"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/184929/files/IPSJ-TOD1004009.pdf","label":"IPSJ-TOD1004009.pdf"},"date":[{"dateType":"Available","dateValue":"2019-12-13"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-TOD1004009.pdf","filesize":[{"value":"579.9 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"13"},{"tax":["include_tax"],"price":"0","billingrole":"39"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"be100dc6-2939-41c2-81f7-e7ecb14af3eb","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2017 by the Information Processing Society of Japan"}]},"item_3_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"木村, 正成"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"若林, 啓"}],"nameIdentifiers":[{}]}]},"item_3_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Masanari, Kimura","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Kei, Wakabayashi","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_3_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA11464847","subitem_source_identifier_type":"NCID"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_6501","resourcetype":"journal article"}]},"item_3_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"1882-7799","subitem_source_identifier_type":"ISSN"}]},"item_3_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"多くの半教師あり学習の目標は，ラベル付きデータとラベルなしデータをうまく組み合わせて分類性能の高いモデルを作ることである．半教師あり学習でよく用いられる手法の1つにラベル伝播があるが，確信度の低いデータにもラベルをつけてしまうため性能が落ちてしまうという問題がある．本研究では，有向グラフ上でのランダムウォークにいくつかのルールを課した新しい手法を提案し，深層学習などのより強力な分類器を学習することによって，ラベル付きデータが極端に少ないケースであっても高い性能のモデルを学習することができる．実験では，提案手法をベンチマークデータに適用し，既存の単純なラベル伝播を行ってラベルを増やした手法と比較してそれを上回る結果が得られた．","subitem_description_type":"Other"}]},"item_3_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"The goal of many semi supervised learning is to make a model with high classification performance by successfully combining labeled data and unlabeled data. Label propagation is one of the techniques often used in semi supervised learning, but there is a problem that performance is lowered because labels are attached to data with low confidence. In this research, we propose a new method that imposes several rules on random walk on directed graph, learning stronger classifiers such as deep learning, and so on, it is a case with extremely few labeled data even with a high performance model you can learn. Experiments show that the proposed method is applied to benchmark data, and compared with the method in which labels are increased by performing existing simple label propagation, the result exceeding that is obtained.","subitem_description_type":"Other"}]},"item_3_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"40","bibliographic_titles":[{"bibliographic_title":"情報処理学会論文誌データベース（TOD）"}],"bibliographicPageStart":"36","bibliographicIssueDates":{"bibliographicIssueDate":"2017-12-13","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"4","bibliographicVolumeNumber":"10"}]},"relation_version_is_last":true,"weko_creator_id":"11"},"id":184929,"updated":"2025-01-20T03:07:59.417239+00:00","links":{},"created":"2025-01-19T00:52:12.258617+00:00"}