Semi-supervised learning with connectivity-driven convolutional neural networks

Página do item simplificado
dc.contributor.authorAmorim, Willian Paraguassu
dc.contributor.authorRosa, Gustavo Henrique [UNESP]
dc.contributor.authorThomazella, Rogério [UNESP]
dc.contributor.authorCastanho, José Eduardo Cogo [UNESP]
dc.contributor.authorDotto, Fábio Romano Lofrano [UNESP]
dc.contributor.authorJúnior, Oswaldo Pons Rodrigues
dc.contributor.authorMarana, Aparecido Nilceu [UNESP]
dc.contributor.authorPapa, João Paulo [UNESP]
dc.contributor.institutionFederal University of Grande Dourados
dc.contributor.institutionUniversidade Estadual Paulista (Unesp)
dc.contributor.institutionCorumbá Concessões S.A
dc.date.accessioned2019-10-06T17:18:08Z
dc.date.available2019-10-06T17:18:08Z
dc.date.issued2019-12-01
dc.description.abstractThe annotation of large datasets is an issue whose challenge increases as the number of labeled samples available to train the classifier reduces in comparison to the amount of unlabeled data. In this context, semi-supervised learning methods aim at discovering and propagating labels to unlabeled samples, such that their correct labeling can improve the classification performance. In this work, we propose a semi-supervised methodology that explores the optimum connectivity among unlabeled samples through the Optimum-Path Forest (OPF) classifier to improve the learning process of Convolution Neural Networks (CNNs). Our proposal makes use of the OPF to classify an unlabeled training set that is used to pre-train a CNN for further fine-tuning using the limited labeled data only. The proposed approach is experimentally validated on traditional datasets and provides competitive results in comparison to state-of-the-art semi-supervised learning methods.en
dc.description.affiliationFederal University of Grande Dourados
dc.description.affiliationSão Paulo State University - UNESP
dc.description.affiliationCorumbá Concessões S.A
dc.description.affiliationUnespSão Paulo State University - UNESP
dc.description.sponsorshipFundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
dc.description.sponsorshipConselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
dc.description.sponsorshipIdFAPESP: 2013/07375-0
dc.description.sponsorshipIdFAPESP: 2014/12236-1
dc.description.sponsorshipIdFAPESP: 2015/25739-4
dc.description.sponsorshipIdFAPESP: 2016/19403-6
dc.description.sponsorshipIdCNPq: 307066/2017-7
dc.description.sponsorshipIdCNPq: 427968/2018-6
dc.format.extent16-22
dc.identifierhttp://dx.doi.org/10.1016/j.patrec.2019.08.012
dc.identifier.citationPattern Recognition Letters, v. 128, p. 16-22.
dc.identifier.doi10.1016/j.patrec.2019.08.012
dc.identifier.issn0167-8655
dc.identifier.scopus2-s2.0-85070863519
dc.identifier.urihttp://hdl.handle.net/11449/190583
dc.language.isoeng
dc.relation.ispartofPattern Recognition Letters
dc.rights.accessRightsAcesso aberto
dc.sourceScopus
dc.subjectConvolutional neural networks
dc.subjectOptimum-path forest
dc.subjectSemi-supervised learning
dc.titleSemi-supervised learning with connectivity-driven convolutional neural networksen
dc.typeArtigo
unesp.author.lattes6027713750942689[7]
unesp.author.orcid0000-0002-2074-5152[3]
unesp.author.orcid0000-0002-6494-7514[8]
unesp.author.orcid0000-0003-4861-7061[7]
unesp.campusUniversidade Estadual Paulista (Unesp), Faculdade de Ciências, Baurupt
unesp.departmentComputação - FCpt

Arquivos

Coleções

Artigos - Engenharia Elétrica - FEB
Bauru - FC - Faculdade de Ciências