{"links":{},"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00234154","sets":["1164:4619:11539:11642"]},"path":["11642"],"owner":"44499","recid":"234154","title":["集合データのための識別と生成のマルチタスク学習"],"pubdate":{"attribute_name":"公開日","attribute_value":"2024-05-08"},"_buckets":{"deposit":"a6126bce-3e29-48b9-87ba-feabdcfd534b"},"_deposit":{"id":"234154","pid":{"type":"depid","value":"234154","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"集合データのための識別と生成のマルチタスク学習","author_link":["637531","637533","637532"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"集合データのための識別と生成のマルチタスク学習"},{"subitem_title":"Multi-Task Learning of Classification and Generation for Set Data","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"卒論スポットライトセッション (CVIM)","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2024-05-08","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"大阪大学"},{"subitem_text_value":"大阪大学"},{"subitem_text_value":"大阪大学"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Osaka Uniersity","subitem_text_language":"en"},{"subitem_text_value":"Osaka Uniersity","subitem_text_language":"en"},{"subitem_text_value":"Osaka Uniersity","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/234154/files/IPSJ-CVIM24238023.pdf","label":"IPSJ-CVIM24238023.pdf"},"date":[{"dateType":"Available","dateValue":"2026-05-08"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-CVIM24238023.pdf","filesize":[{"value":"1.4 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"20"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"b899baa0-a32b-4d37-9e61-aebeb77b8b37","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2024 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"佐藤, 史興"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"早志, 英朗"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"長原, 一"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA11131797","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":"2188-8701","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"本研究では，集合データを対象とした識別と生成のマルチタスク学習モデルを提案する．提案モデルでは，集合データの特性である不変性や同変性を保ちながら変分オートエンコーダと識別層を構成することで，データの生成と識別を単一のネットワークで学習する．これにより，集合データ識別において半教師あり学習が可能となる．さらに，生成モデルにより推定した入力データ分布に基づく信頼度較正にも応用できる．実験では，半教師あり識別タスクにおける提案モデルの性能を評価するとともに，識別器のみで構成されるモデルと比較した．その結果，識別と生成の同時学習が集合データの識別精度向上と信頼度較正に有効であり，未ラベルデータの割合が高い場合にも効果的であることが確認された．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"In this study, we propose a multi-task learning model of classification and generation for set data. The proposed model learns data generation and classification in a single neural network by integrating a classification layer into a variational autoencoder while maintaining permutation invariance and equivariance nature, which are characteristics of set data. This enables semi-supervised learning in set data classification. It can also be applied to confidence calibration using the input data distribution estimated by the generative model. In the experiments, we evaluated the performance of the proposed model in a semi-supervised classification task and compared it with a model consisting only of a classifier. The results showed that simultaneous learning of the classification and generation tasks is effective in improving the classification accuracy and confidence calibration for set data, even when the proportion of unlabeled data is high.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告コンピュータビジョンとイメージメディア（CVIM）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2024-05-08","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"23","bibliographicVolumeNumber":"2024-CVIM-238"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"created":"2025-01-19T01:35:49.698751+00:00","updated":"2025-01-19T09:53:04.872453+00:00","id":234154}