Naive semi-supervised deep learning using pseudo-label

Cited 22 time in webofscience Cited 21 time in scopus
  • Hit : 665
  • Download : 0
DC FieldValueLanguage
dc.contributor.authorLi, Zhunko
dc.contributor.authorKo, ByungSooko
dc.contributor.authorChoi, Ho-Jinko
dc.date.accessioned2019-10-01T00:20:04Z-
dc.date.available2019-10-01T00:20:04Z-
dc.date.created2019-07-01-
dc.date.issued2019-09-
dc.identifier.citationPEER-TO-PEER NETWORKING AND APPLICATIONS, v.12, no.5, pp.1358 - 1368-
dc.identifier.issn1936-6442-
dc.identifier.urihttp://hdl.handle.net/10203/267705-
dc.description.abstractTo facilitate the utilization of large-scale unlabeled data, we propose a simple and effective method for semi-supervised deep learning that improves upon the performance of the deep learning model. First, we train a classifier and use its outputs on unlabeled data as pseudo-labels. Then, we pre-train the deep learning model with the pseudo-labeled data and fine-tune it with the labeled data. The repetition of pseudo-labeling, pre-training, and fine-tuning is called naive semi-supervised deep learning. We apply this method to the MNIST, CIFAR-10, and IMDB data sets, which are each divided into a small labeled data set and a large unlabeled data set by us. Our method achieves significant performance improvements compared to the deep learning model without pre-training. We further analyze the factors that affect our method to provide a better understanding of how to utilize naive semi-supervised deep learning in practical application.-
dc.languageEnglish-
dc.publisherSPRINGER-
dc.titleNaive semi-supervised deep learning using pseudo-label-
dc.typeArticle-
dc.identifier.wosid000484477700024-
dc.identifier.scopusid2-s2.0-85058144754-
dc.type.rimsART-
dc.citation.volume12-
dc.citation.issue5-
dc.citation.beginningpage1358-
dc.citation.endingpage1368-
dc.citation.publicationnamePEER-TO-PEER NETWORKING AND APPLICATIONS-
dc.identifier.doi10.1007/s12083-018-0702-9-
dc.contributor.localauthorChoi, Ho-Jin-
dc.contributor.nonIdAuthorLi, Zhun-
dc.contributor.nonIdAuthorKo, ByungSoo-
dc.description.isOpenAccessN-
dc.type.journalArticleArticle-
dc.subject.keywordAuthorDeep learning-
dc.subject.keywordAuthorSemi-supervised learning-
dc.subject.keywordAuthorPseudo-label-
dc.subject.keywordAuthorPre-training-
Appears in Collection
CS-Journal Papers(저널논문)
Files in This Item
There are no files associated with this item.
This item is cited by other documents in WoS
⊙ Detail Information in WoSⓡ Click to see webofscience_button
⊙ Cited 22 items in WoS Click to see citing articles in records_button

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0