{"created":"2025-01-19T01:12:50.646535+00:00","updated":"2025-01-19T17:42:38.421227+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00211704","sets":["1164:5352:10544:10612"]},"path":["10612"],"owner":"44499","recid":"211704","title":["Nonparametric Bayesian Deep Visualization"],"pubdate":{"attribute_name":"公開日","attribute_value":"2021-06-21"},"_buckets":{"deposit":"d196d952-ad4a-4ba1-8853-37c5f4b42c29"},"_deposit":{"id":"211704","pid":{"type":"depid","value":"211704","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"Nonparametric Bayesian Deep Visualization","author_link":["538268","538270","538269","538267"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"Nonparametric Bayesian Deep Visualization"},{"subitem_title":"Nonparametric Bayesian Deep Visualization","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"深層学習・行列分解","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2021-06-21","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"株式会社ブリヂストン"},{"subitem_text_value":"統計数理研究所"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Bridgestone Corporation","subitem_text_language":"en"},{"subitem_text_value":"The Institute of Statistical Mathematics ","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/211704/files/IPSJ-BIO21066001.pdf","label":"IPSJ-BIO21066001.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-BIO21066001.pdf","filesize":[{"value":"3.2 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"41"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"fcde5b6e-0816-435c-b8b2-d64ddb1c3c5d","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2021 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":"石塚, 治也"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"持橋, 大地"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Haruya, Ishizuka","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Daichi, Mochihashi","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AA12055912","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":"2188-8590","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"高次元データを散布図で可視化する場合,次元削減により観測値を圧縮する必要がある.t-SNE に代表される,観測値間の類似度を元に次元削減を行う類似度ベース次元削減は,可視化の際に広く用いられている.しかし, 観測値のベクトル表現によっては,観測値間の類似度と真の類似度が乖離するため,可視化精度が低下する.これに対して,ニューラルネットワーク (NN) を用いる深層潜在変数モデルは,観測値ベクトルよりも正確に特徴を反映する潜在表現を推定ができる可能性があり,ベクトル表現が原因で前者が機能しないときの有効な代替案になる.一方で性能の最適化には,NN のモデル構造など多くの超パラメータを調整する必要があり,その試行の中で,NN の大量のパラメータの学習を繰り返すため,計算時間が増大しやすい.また,可視化結果は設定された超パラメータの探索範囲によって変化する.本稿では,これらの問題点に対処するため,Nonparametric Bayesian Deep Visualization (NPDV) を提案する.NPDV は,NN による潜在表現の推定と可視化を同時に行う確率モデルであり,無限混合ガウスモデル,無限ユニット NN を併用することで,少数の超パラメータでモデルが構成される.さらに,無限ユニット NN は少数のパラメータで定義されるため,パラメータ数も既存の深層潜在変数モデルと比較して少ない.本稿では提案手法の詳細と,実験結果について報告する.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告バイオ情報学(BIO)"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2021-06-21","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"1","bibliographicVolumeNumber":"2021-BIO-66"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":211704,"links":{}}