Semi-Supervised Classification Based on Classification from Positive and Unlabeled Data

Most of the semi-supervised classification methods developed so far use unlabeled data for regularization purposes under particular distributional assumptions such as the cluster assumption. In contrast, recently developed methods of classification from positive and unlabeled data (PU classification...

Full description

Saved in:
Bibliographic Details
Main Authors Sakai, Tomoya, Plessis, Marthinus Christoffel du, Niu, Gang, Sugiyama, Masashi
Format Journal Article
LanguageEnglish
Published 23.05.2016
Subjects
Computer Science - Learning
Online AccessGet full text

Cover

Abstract Most of the semi-supervised classification methods developed so far use unlabeled data for regularization purposes under particular distributional assumptions such as the cluster assumption. In contrast, recently developed methods of classification from positive and unlabeled data (PU classification) use unlabeled data for risk evaluation, i.e., label information is directly extracted from unlabeled data. In this paper, we extend PU classification to also incorporate negative data and propose a novel semi-supervised classification approach. We establish generalization error bounds for our novel methods and show that the bounds decrease with respect to the number of unlabeled data without the distributional assumptions that are required in existing semi-supervised classification methods. Through experiments, we demonstrate the usefulness of the proposed methods.
AbstractList Most of the semi-supervised classification methods developed so far use unlabeled data for regularization purposes under particular distributional assumptions such as the cluster assumption. In contrast, recently developed methods of classification from positive and unlabeled data (PU classification) use unlabeled data for risk evaluation, i.e., label information is directly extracted from unlabeled data. In this paper, we extend PU classification to also incorporate negative data and propose a novel semi-supervised classification approach. We establish generalization error bounds for our novel methods and show that the bounds decrease with respect to the number of unlabeled data without the distributional assumptions that are required in existing semi-supervised classification methods. Through experiments, we demonstrate the usefulness of the proposed methods.
Author Sugiyama, Masashi
Niu, Gang
Plessis, Marthinus Christoffel du
Sakai, Tomoya
Author_xml – sequence: 1
  givenname: Tomoya
  surname: Sakai
  fullname: Sakai, Tomoya
– sequence: 2
  givenname: Marthinus Christoffel du
  surname: Plessis
  fullname: Plessis, Marthinus Christoffel du
– sequence: 3
  givenname: Gang
  surname: Niu
  fullname: Niu, Gang
– sequence: 4
  givenname: Masashi
  surname: Sugiyama
  fullname: Sugiyama, Masashi
BackLink https://doi.org/10.48550/arXiv.1605.06955$$DView paper in arXiv
BookMark eNpdj09LxDAUxHPQg65-AE_mC7SmTV-SPWr9CwsKu-KxvDQvEGjTJa1Fv73d1ZOnGYZhht85O4lDJMauCpFXBkDcYPoKc14oAblQa4Az9rGlPmTbzz2lOYzkeN3hOAYfWpzCEPkdHsLF_Mt9Gnr-NoxhCjNxjI6_xw4tdUv7Hie8YKceu5Eu_3TFdo8Pu_o527w-vdS3mwyVhqzUXqHxALZABU4uVhIJo5VWtvRSlMZBS61dV9aAraRCXUjhvVyuWlfJFbv-nT2SNfsUekzfzYGwORLKH6KaTmo
ContentType Journal Article
Copyright http://arxiv.org/licenses/nonexclusive-distrib/1.0
Copyright_xml – notice: http://arxiv.org/licenses/nonexclusive-distrib/1.0
DBID AKY
GOX
DOI 10.48550/arxiv.1605.06955
DatabaseName arXiv Computer Science
arXiv.org
DatabaseTitleList
Database_xml – sequence: 1
  dbid: GOX
  name: arXiv.org
  url: http://arxiv.org/find
  sourceTypes: Open Access Repository
DeliveryMethod fulltext_linktorsrc
ExternalDocumentID 1605_06955
GroupedDBID AKY
GOX
ID FETCH-LOGICAL-a675-27f6a8f55b1a65d38f53ee087676b2f3028d5cecb94b85b436a7130ff3abecd43
IEDL.DBID GOX
IngestDate Mon Jan 08 05:45:32 EST 2024
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed false
IsScholarly false
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a675-27f6a8f55b1a65d38f53ee087676b2f3028d5cecb94b85b436a7130ff3abecd43
OpenAccessLink https://arxiv.org/abs/1605.06955
ParticipantIDs arxiv_primary_1605_06955
PublicationCentury 2000
PublicationDate 2016-05-23
PublicationDateYYYYMMDD 2016-05-23
PublicationDate_xml – month: 05
  year: 2016
  text: 2016-05-23
  day: 23
PublicationDecade 2010
PublicationYear 2016
Score 1.6322405
SecondaryResourceType preprint
Snippet Most of the semi-supervised classification methods developed so far use unlabeled data for regularization purposes under particular distributional assumptions...
SourceID arxiv
SourceType Open Access Repository
SubjectTerms Computer Science - Learning
Title Semi-Supervised Classification Based on Classification from Positive and Unlabeled Data
URI https://arxiv.org/abs/1605.06955
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwdV1LSwMxEB7anryIolKf5OA1aDeP3R7roxbBB7TFvS2ZTQI9uJZ2W_z5TrIriuAtTB6HLyTfN8lkAnBJOx6pBnQcM9Rc-lRxJJbgPkXib6OyDMM55NOznszlY67yDrDvtzBm9bnYNvmBcX010OHIQw-V6kI3SULI1sNL3lxOxlRcbfufdqQxo-kXSYz3YLdVd2zUTMc-dFx1AG9T977g080yLMu1syx-RBlCdCIq7MYEIxX-2MPLD_Yag6q2jpHHz-YVzRnxhGV3pjaHMBvfz24nvP3PgBuS5TxJvTaZVwoHRisrqCicCynhUo2JF8T0VpWuxKHETKEU2pAHee29oKFLK8UR9KqPyvWBGZkaq6gzMYskpwKFKIMQUmicLm12DP2IQrFsUlYUAaAiAnTyf9Up7JAc0OFuPBFn0KtXG3dOlFvjRcT9C4Z0geU
link.rule.ids 228,230,783,888
linkProvider Cornell University
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Semi-Supervised+Classification+Based+on+Classification+from+Positive+and+Unlabeled+Data&rft.au=Sakai%2C+Tomoya&rft.au=Plessis%2C+Marthinus+Christoffel+du&rft.au=Niu%2C+Gang&rft.au=Sugiyama%2C+Masashi&rft.date=2016-05-23&rft_id=info:doi/10.48550%2Farxiv.1605.06955&rft.externalDocID=1605_06955