{"id":198998,"updated":"2025-01-19T21:44:49.605882+00:00","links":{},"created":"2025-01-19T01:03:07.411258+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00198998","sets":["1164:4619:9659:9887"]},"path":["9887"],"owner":"44499","recid":"198998","title":["モデル選択によるニューラルネットワークの簡易的ドメイン適応"],"pubdate":{"attribute_name":"公開日","attribute_value":"2019-08-28"},"_buckets":{"deposit":"c343da6e-9ac8-48cf-9796-fe180867250e"},"_deposit":{"id":"198998","pid":{"type":"depid","value":"198998","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"モデル選択によるニューラルネットワークの簡易的ドメイン適応","author_link":["481112","481111","481109","481110","481108","481113"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"モデル選択によるニューラルネットワークの簡易的ドメイン適応"},{"subitem_title":"Rough Domain Adaptation through Model Selection for Neural Networks","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"セッション5","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2019-08-28","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"日本電気株式会社"},{"subitem_text_value":"日本電気株式会社"},{"subitem_text_value":"日本電気株式会社"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"NEC Corporation","subitem_text_language":"en"},{"subitem_text_value":"NEC Corporation","subitem_text_language":"en"},{"subitem_text_value":"NEC Corporation","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":44499,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/198998/files/IPSJ-CVIM19218022.pdf","label":"IPSJ-CVIM19218022.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-CVIM19218022.pdf","filesize":[{"value":"493.9 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"20"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"e73f34e0-cf32-42ef-a615-6ab4c19833dd","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2019 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":"澤田, あずさ"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"柴田, 剛志"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"谷内田, 尚司"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Azusa, Sawada","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Takashi, Shibata","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Shoji, Yachida","creatorNameLang":"en"}],"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":"機械学習により得られた認識モデルは，入力データ分布が学習時とずれるドメインシフトの下で性能が劣化する．一般には既知のシフトを学習に用いるなどして頑健なモデルを作成することが多いが，ピーク性能が落ちる場合がある．そこで本研究では万能な単一モデルを作る代わりに，学習に用いるデータ拡張方法の違いで作った複数のモデル候補を作成しておき，そこからの選択により未知のシフトに簡易的に適応することを考える．候補に分割して選択することでピーク性能の改善が見込めることを，数字認識，VisDa 分類データを用いた単一ソースドメインからの学習実験で確認した．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"The performance of recognition models by machine learning degrades under data domain shifts between training and test phase. They usually train robust models against known shifts, which may lead to marginal performances in some cases. We propose to adapt roughly to unseen shifts by model selection from candidates trained under different data augmentation functions instead of making a single robust model. Our Experiments with single source domain show that the selection from separated candidates can improve test performance using digit recognition datasets and VisDa classification dataset.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"5","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":"2019-08-28","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"22","bibliographicVolumeNumber":"2019-CVIM-218"}]},"relation_version_is_last":true,"weko_creator_id":"44499"}}