{"created":"2025-01-19T01:12:10.597977+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00210981","sets":["1164:2592:10486:10582"]},"path":["10582"],"owner":"44499","recid":"210981","title":["On Learning from Average-Case Errorless Computing (Extended Abstract)"],"pubdate":{"attribute_name":"公開日","attribute_value":"2021-04-30"},"_buckets":{"deposit":"e49385f0-b1d6-4834-b58d-eceba64d7161"},"_deposit":{"id":"210981","pid":{"type":"depid","value":"210981","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"On Learning from Average-Case Errorless Computing (Extended Abstract)","author_link":["535206","535205"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"On Learning from Average-Case Errorless Computing (Extended Abstract)"},{"subitem_title":"On Learning from Average-Case Errorless Computing (Extended Abstract)","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2021-04-30","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"Department of Mathematical and Computing Science, Tokyo Institute of Technology "}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Department of Mathematical and Computing Science, Tokyo Institute of Technology","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/210981/files/IPSJ-AL21183005.pdf","label":"IPSJ-AL21183005.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-AL21183005.pdf","filesize":[{"value":"815.8 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"9"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"a8bf2eff-1c6d-4efe-a354-19ef87348386","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":"Mikito, Nanashima"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Mikito, Nanashima","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN1009593X","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":"2188-8566","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"A PAC learning model contains a worst-case sense due to the following requirement : a learner must learn (1) all functions in a concept class on (2) all example distributions. Thus, whether average-case computation on a fixed distribution is sufficient for such learning is quite non-trivial and a significant question to understand the nature of learning. Recent studies on complexity theory implicitly revealed that average-case errorless computation for a distributional NP-problem is sufficient to resolve the worst-case requirement (1) alone. In this paper, we addressed the worst-case requirement (2) (alone) and the requirements (1) and (2) simultaneously to identify where the current non-average aspect of learning essentially arises. Specifically, we will show the following theorems: (i) polynomial-size circuits are efficiently distribution-free PAC learnable on average iff there exist auxiliary-input pseudorandom generators, and (ii) if DistNP ⊆ AvgP, then polynomial-size circuits are efficiently agnostic learnable under all P/poly-computable example distributions. Our learning algorithm works without specified an example distribution as a usual distribution-free learner, but the time and sample complexity depends on the complexity of example distributions. ","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"A PAC learning model contains a worst-case sense due to the following requirement : a learner must learn (1) all functions in a concept class on (2) all example distributions. Thus, whether average-case computation on a fixed distribution is sufficient for such learning is quite non-trivial and a significant question to understand the nature of learning. Recent studies on complexity theory implicitly revealed that average-case errorless computation for a distributional NP-problem is sufficient to resolve the worst-case requirement (1) alone. In this paper, we addressed the worst-case requirement (2) (alone) and the requirements (1) and (2) simultaneously to identify where the current non-average aspect of learning essentially arises. Specifically, we will show the following theorems: (i) polynomial-size circuits are efficiently distribution-free PAC learnable on average iff there exist auxiliary-input pseudorandom generators, and (ii) if DistNP ⊆ AvgP, then polynomial-size circuits are efficiently agnostic learnable under all P/poly-computable example distributions. Our learning algorithm works without specified an example distribution as a usual distribution-free learner, but the time and sample complexity depends on the complexity of example distributions. ","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"8","bibliographic_titles":[{"bibliographic_title":"研究報告アルゴリズム（AL）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2021-04-30","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"5","bibliographicVolumeNumber":"2021-AL-183"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":210981,"updated":"2025-01-19T17:57:19.235888+00:00","links":{}}