{"id":189700,"updated":"2025-01-20T01:30:08.557720+00:00","links":{},"created":"2025-01-19T00:55:42.909293+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00189700","sets":["1164:5352:9434:9497"]},"path":["9497"],"owner":"11","recid":"189700","title":["CapsuleNet を用いた半教師ありクラスタリングによる未知ラベルの検出"],"pubdate":{"attribute_name":"公開日","attribute_value":"2018-06-06"},"_buckets":{"deposit":"da826199-627b-47e8-a2cb-61d35b66e937"},"_deposit":{"id":"189700","pid":{"type":"depid","value":"189700","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"CapsuleNet を用いた半教師ありクラスタリングによる未知ラベルの検出","author_link":["432461","432462","432460","432463"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"CapsuleNet を用いた半教師ありクラスタリングによる未知ラベルの検出"},{"subitem_title":"Semi-supervised Clustering for Detecting Unlabeled Class using CapsuleNet","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"IBISML一般セッション","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2018-06-06","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京大学大学院工学系研究科システム創成学専攻"},{"subitem_text_value":"東京大学大学院工学系研究科システム創成学専攻"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"the University of Tokyo School of Enginnering Department of Systems Innovation","subitem_text_language":"en"},{"subitem_text_value":"the University of Tokyo School of Enginnering Department of Systems Innovation","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/189700/files/IPSJ-BIO18054007.pdf","label":"IPSJ-BIO18054007.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-BIO18054007.pdf","filesize":[{"value":"652.2 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"41"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"6205f1fa-4ce1-4bb5-8027-31cb1cbccf0d","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2018 by the Institute of Electronics, Information and Communication Engineers This SIG report is only available to those in membership of the SIG."}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"福馬, 智生"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"鳥海, 不二夫"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Fukuma, Tomoki","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Toriumi, Fujio","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA12055912","subitem_source_identifier_type":"NCID"}]},"item_4_textarea_12":{"attribute_name":"Notice","attribute_value_mlt":[{"subitem_textarea_value":"SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc."}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_18gh","resourcetype":"technical report"}]},"item_4_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2188-8590","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"推論を行う際かつて学習の際用意していたラベルセットのどれにも属さないようなラベルのデータが入力された場合，それらを未知と検出できることは大変重要である．本研究ではラベル付きのデータセットとラベルが付与されていないデータセットを同時に学習し，既存のクラスのどれにも属さないような未知ラベルの検出を特徴抽出の過程から End-to-End で行う半教師あり手法を提案する．MNIST データセットを使った精度検証において，全 10 個の学習ラベルのうち 5 個のみを学習に用いて学習を行った際，テストデータについて未知ラベルかどうかの二値分類で約 80%，10 クラスへのクラスタリングで約 82% の精度となり，従来手法よりも遥かに高い精度で未知クラスかどうかの判別と複数個の発見が可能であることが示された．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告バイオ情報学（BIO）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2018-06-06","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"7","bibliographicVolumeNumber":"2018-BIO-54"}]},"relation_version_is_last":true,"weko_creator_id":"11"}}