{"links":{},"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00225946","sets":["1164:4619:11188:11274"]},"path":["11274"],"owner":"44499","recid":"225946","title":["Towards Better Representation and Interpretability for Deep Neural Networks on Visual Tasks"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-05-11"},"_buckets":{"deposit":"bd3dc7a5-24ce-441f-85ac-83df3f9926a8"},"_deposit":{"id":"225946","pid":{"type":"depid","value":"225946","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"Towards Better Representation and Interpretability for Deep Neural Networks on Visual Tasks","author_link":["599066","599069","599070","599068","599071","599067","599065","599064"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"Towards Better Representation and Interpretability for Deep Neural Networks on Visual Tasks"},{"subitem_title":"Towards Better Representation and Interpretability for Deep Neural Networks on Visual Tasks","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"D論セッション","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2023-05-11","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"Osaka University"},{"subitem_text_value":"Osaka University"},{"subitem_text_value":"Osaka University"},{"subitem_text_value":"Osaka University"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Osaka University","subitem_text_language":"en"},{"subitem_text_value":"Osaka University","subitem_text_language":"en"},{"subitem_text_value":"Osaka University","subitem_text_language":"en"},{"subitem_text_value":"Osaka University","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"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/225946/files/IPSJ-CVIM23234002.pdf","label":"IPSJ-CVIM23234002.pdf"},"date":[{"dateType":"Available","dateValue":"2025-05-11"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-CVIM23234002.pdf","filesize":[{"value":"3.4 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":"747ab99d-d278-40f4-a090-1cd5c4695fcc","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2023 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Bowen, Wang"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Liangzhi, Li"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yuta, Nakashima"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Hajime, Nagahara"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Bowen, Wang","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Liangzhi, Li","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yuta, Nakashima","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Hajime, Nagahara","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":"Deep Neural Networks (DNNs) have shown their power in many research fields, and related applications are entering people's daily lives with unstoppable momentum. However, the large number of DNNs' training parameters causes difficulty in learning representation from real-world data efficiently, and the black-box nature harms its explainability. In this thesis, we will show how to design a DNN for better representation, as well as interpret its behavior for reliable artificial intelligence (AI). By embedding a slot-attention-based XAI module, we find that a DNN model is interpretable, and the learning of representation can be benefited from this interpretability. XAI methods are further extended to find representation in a simple classification task. The found representation is transferred as training data for a complex object detection task, realizing weak supervision. In two different real-world scenarios, we evaluate that our proposal can encourage DNNs to learn better representation and let them be interpretable.","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Deep Neural Networks (DNNs) have shown their power in many research fields, and related applications are entering people's daily lives with unstoppable momentum. However, the large number of DNNs' training parameters causes difficulty in learning representation from real-world data efficiently, and the black-box nature harms its explainability. In this thesis, we will show how to design a DNN for better representation, as well as interpret its behavior for reliable artificial intelligence (AI). By embedding a slot-attention-based XAI module, we find that a DNN model is interpretable, and the learning of representation can be benefited from this interpretability. XAI methods are further extended to find representation in a simple classification task. The found representation is transferred as training data for a complex object detection task, realizing weak supervision. In two different real-world scenarios, we evaluate that our proposal can encourage DNNs to learn better representation and let them be interpretable.","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":"2023-05-11","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"2","bibliographicVolumeNumber":"2023-CVIM-234"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"created":"2025-01-19T01:25:26.334476+00:00","updated":"2025-01-19T12:38:01.838916+00:00","id":225946}