{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00224945","sets":["1164:4088:11174:11243"]},"path":["11243"],"owner":"44499","recid":"224945","title":["非中央集権型向けのモデル更新履歴を用いた連合学習手法および実用的なフレームワークの設計"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-03-08"},"_buckets":{"deposit":"3d818b47-f775-48be-a961-aa189ce51e54"},"_deposit":{"id":"224945","pid":{"type":"depid","value":"224945","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"非中央集権型向けのモデル更新履歴を用いた連合学習手法および実用的なフレームワークの設計","author_link":["594365","594369","594367","594368","594366"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"非中央集権型向けのモデル更新履歴を用いた連合学習手法および実用的なフレームワークの設計"},{"subitem_title":"A federated learning using model update history and a design of a practical framework for decentralized systems","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"IOTセッション","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2023-03-08","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京大学"},{"subitem_text_value":"東京大学"},{"subitem_text_value":"東京大学"},{"subitem_text_value":"東京大学"},{"subitem_text_value":"東京大学"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"The University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"The University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"The University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"The University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"The University of Tokyo","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/224945/files/IPSJ-IOT23060005.pdf","label":"IPSJ-IOT23060005.pdf"},"date":[{"dateType":"Available","dateValue":"2025-03-08"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-IOT23060005.pdf","filesize":[{"value":"1.3 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":"43"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"20f15793-2097-4f74-afd9-48762fe8f5dd","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":"金森, 勇介"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"山﨑, 雄輔"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"細合, 晋太郎"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"中村, 宏"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"高瀬, 英希"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA12326962","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":"2188-8787","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"プライバシー保護の観点から，個人情報など元データはやりとりせずに各デバイスからのモデルを統合する機械学習技術である連合学習に注目が集まっている．本研究では，非同期にモデルが統合されデバイスの接続構造が非中央集権型である連合学習について，効率的かつ実用的な技術の確立を目指す．まず，モデルの更新履歴を用いた連合学習手法を提案する．各ラウンドのモデルの統合段階において，前回のモデル統合時と比べて更新があったモデルのみを用いてモデルを統合する．これによって，古いモデルとの統合による学習の遅れを防ぎ，より効率的なモデルの統合が期待される．また，非同期式・非中央集権型の連合学習においては学習と統合の順序が任意であることに着目し，順序の違いが学習に及ぼす影響を考慮したアルゴリズムを提案する．次に，実機への応用性で強みを持つ Flower を拡張し，非同期式・非中央集権型の連合学習に対応した実用的なフレームワークを設計する．Flower が採用している gRPC を用いて効率的な通信が実現でき，中央サーバにあたるものを配置せずに実機への拡張ができることが期待される．提案手法についての評価を行い，更新履歴を用いたモデル統合が有効であることを示した．統合と学習の順序については，条件設定に応じて優位となる順序が異なることを示した．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告インターネットと運用技術（IOT）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2023-03-08","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"5","bibliographicVolumeNumber":"2023-IOT-60"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"updated":"2025-01-19T12:58:47.747104+00:00","created":"2025-01-19T01:24:27.835963+00:00","links":{},"id":224945}