{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00241817","sets":["1164:3027:11889:11890"]},"path":["11890"],"owner":"44499","recid":"241817","title":["CNNを用いた絵画推薦システムの精度向上に関する検討"],"pubdate":{"attribute_name":"公開日","attribute_value":"2025-01-07"},"_buckets":{"deposit":"4fa116b2-fb83-4ba2-ad23-0833b7dadbf1"},"_deposit":{"id":"241817","pid":{"type":"depid","value":"241817","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"CNNを用いた絵画推薦システムの精度向上に関する検討","author_link":["666355","666357","666358","666356"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"CNNを用いた絵画推薦システムの精度向上に関する検討"},{"subitem_title":"Study on Improving the Accuracy of a Painting Recommendation System Using CNN","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"セッション1: 機械学習・大規模言語モデル","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2025-01-07","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"日本大学大学院生産工学研究科"},{"subitem_text_value":"日本大学大学院生産工学研究科"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Graduate School of Industrial Technology, Nihon University","subitem_text_language":"en"},{"subitem_text_value":"Graduate School of Industrial Technology, Nihon University","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/241817/files/IPSJ-HCI25211005.pdf","label":"IPSJ-HCI25211005.pdf"},"date":[{"dateType":"Available","dateValue":"2027-01-07"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-HCI25211005.pdf","filesize":[{"value":"2.1 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":"33"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"b1735087-89f0-486e-8f88-61683e935d49","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2025 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"深見, 太一"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"中村, 喜宏"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Taichi, Fukami","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yoshihiro, Nakamura","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA1221543X","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":"2188-8760","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"現代社会において，多くの人が絵画を楽しんでいるが，膨大な数の作品の中から自分の好みに合う絵画を見つけることは困難である．本研究では，個々のユーザの嗜好を反映する推薦システムを提案する．さまざまな絵画を収集し，実験協力者に評価してもらい，その評価に基づいてユーザを嗜好別にクラスタリングした．その後，畳み込みニューラルネットワーク（CNN）を用いて推薦システムを構築した．結果として，全体を分析するよりも，嗜好によってユーザをクラスタリングした方が高い精度を示すことが確認された．しかし，収集できる学習データの量には限界があるという制約があった．そこで，学習データに AI 生成画像を組み込む可能性を検討した．その結果，人間が作成した絵画を参考に生成された AI 画像を学習データに加えることで，精度の向上が見られた．このことから，ユーザを嗜好でクラスタリングし，さらに CNN の学習データに AI 生成画像を補足することが，推薦精度を向上させるために有効であるかどうか検討した．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"In modern society, many people enjoy paintings; however, finding preferred artworks among a vast number of options can be challenging. This study proposes a recommendation system that reflects individual user preferences. We collected various paintings and had experimental participants rate them, using these ratings to cluster users based on preferences. A recommendation system was then developed using Convolutional Neural Networks (CNN). Results indicated that clustering users by preferences yielded higher accuracy than analyzing the group as a whole. However, a limitation was the small quantity of training data available. To address this, we examined the potential of incorporating AI-generated images into the training data. It was found that including AI-generated images, designed based on human-created artwork, led to improved accuracy. Thus, clustering users by preferences and supplementing CNN training data with AI-generated images were both confirmed to enhance recommendation accuracy.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"7","bibliographic_titles":[{"bibliographic_title":"研究報告ヒューマンコンピュータインタラクション（HCI）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2025-01-07","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"5","bibliographicVolumeNumber":"2025-HCI-211"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":241817,"updated":"2025-01-19T07:31:51.615624+00:00","links":{},"created":"2025-01-19T01:46:39.077409+00:00"}