{"id":234132,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00234132","sets":["1164:4619:11539:11642"]},"path":["11642"],"owner":"44499","recid":"234132","title":["深層学習を用いた偏りのあるデータに対して頑健な学習手法に関する研究"],"pubdate":{"attribute_name":"公開日","attribute_value":"2024-05-08"},"_buckets":{"deposit":"145e26d6-68d2-4493-88b1-8c77e313ead6"},"_deposit":{"id":"234132","pid":{"type":"depid","value":"234132","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"深層学習を用いた偏りのあるデータに対して頑健な学習手法に関する研究","author_link":["637451","637452"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"深層学習を用いた偏りのあるデータに対して頑健な学習手法に関する研究"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"D論セッション (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":"名城大学大学院／現在，株式会社センスタイムジャパン"},{"subitem_text_value":"名城大学"}]},"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/234132/files/IPSJ-CVIM24238001.pdf","label":"IPSJ-CVIM24238001.pdf"},"date":[{"dateType":"Available","dateValue":"2026-05-08"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-CVIM24238001.pdf","filesize":[{"value":"4.8 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":"7edc5ba6-408a-4dee-8e14-e03762ea9d6a","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":"加藤, 聡太"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"堀田, 一弘"}],"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":"本研究では実世界の偏りのあるデータに対して有効な学習方法の研究に取り組む．実世界の環境で無作為に収集されたデータセットには，多くの場合「偏り」が発生する．画像分類問題においては，クラス間のデータ数の割合が不均衡なデータセットになる場合があり，またセマンティックセグメンテーションでは，画像内の検出したい領域が複数ある場合，各領域の大きさに「偏り」が発生しているデータセットがほとんどである．そこで本研究では，(1) クラス不均衡な画像分類，(2) 領域不均衡なセマンティックセグメンテーションの 2 つの観点から，実世界のデータで発生する様々な偏りに対して頑健な，新たな損失関数を提案する．様々なデータセットを用いた評価実験から，提案手法を用いることで従来の手法よりも大幅に精度が改善することを確認した．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"16","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":"1","bibliographicVolumeNumber":"2024-CVIM-238"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"updated":"2025-01-19T09:53:29.101243+00:00","created":"2025-01-19T01:35:47.603641+00:00","links":{}}