Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented]
Autor(a) principal: | |
---|---|
Data de Publicação: | 2023 |
Outros Autores: | , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1016/j.eswa.2022.118995 http://hdl.handle.net/11449/247785 |
Resumo: | Image classification is a critical topic due to its wide application and several challenges associated. Despite the huge progress made last decades, there is still a demand for context-aware image representation approaches capable of taking into the dataset manifold for improving classification accuracy. In this work, a representation learning approach is proposed, based on a novel feature augmentation strategy. The proposed method aims to exploit available contextual similarity information through rank-based manifold learning used to define and assign weights to samples used in augmentation. The approach is validated using CNN-based features and LSTM models to achieve even higher accuracy results on image classification tasks. Experimental results show that the feature augmentation strategy can indeed improve the accuracy of results on widely used image datasets (CIFAR10, Stanford Dogs, Linnaeus5, Flowers102 and Flowers17) in different CNNs (ResNet152, VGG16, DPN92). The results indicate gains up to 20% and show the potential of the developed approach in achieving higher accuracy results for image classification. |
id |
UNSP_7c50855ae4908db6d102d0cec0fe27d6 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/247785 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented]Feature augmentationImage classificationLSTMManifold learningRankingImage classification is a critical topic due to its wide application and several challenges associated. Despite the huge progress made last decades, there is still a demand for context-aware image representation approaches capable of taking into the dataset manifold for improving classification accuracy. In this work, a representation learning approach is proposed, based on a novel feature augmentation strategy. The proposed method aims to exploit available contextual similarity information through rank-based manifold learning used to define and assign weights to samples used in augmentation. The approach is validated using CNN-based features and LSTM models to achieve even higher accuracy results on image classification tasks. Experimental results show that the feature augmentation strategy can indeed improve the accuracy of results on widely used image datasets (CIFAR10, Stanford Dogs, Linnaeus5, Flowers102 and Flowers17) in different CNNs (ResNet152, VGG16, DPN92). The results indicate gains up to 20% and show the potential of the developed approach in achieving higher accuracy results for image classification.Microsoft ResearchFundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)Department of Statistics Applied Mathematics and Computing (DEMAC) São Paulo State University (UNESP), Ave. 24A, 1515, São PauloDepartment of Statistics Applied Mathematics and Computing (DEMAC) São Paulo State University (UNESP), Ave. 24A, 1515, São PauloFAPESP: #2017/25908-6FAPESP: #2018/15597-6FAPESP: #2020/02183-9FAPESP: #2020/11366-0CNPq: #309439/2020-5CNPq: #422667/2021-8Universidade Estadual Paulista (UNESP)Pereira-Ferrero, Vanessa Helena [UNESP]Valem, Lucas Pascotti [UNESP]Pedronette, Daniel Carlos Guimarães [UNESP]2023-07-29T13:25:50Z2023-07-29T13:25:50Z2023-03-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/articlehttp://dx.doi.org/10.1016/j.eswa.2022.118995Expert Systems with Applications, v. 213.0957-4174http://hdl.handle.net/11449/24778510.1016/j.eswa.2022.1189952-s2.0-85140439236Scopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengExpert Systems with Applicationsinfo:eu-repo/semantics/openAccess2023-07-29T13:25:50Zoai:repositorio.unesp.br:11449/247785Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T13:56:32.835680Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
title |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
spellingShingle |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] Pereira-Ferrero, Vanessa Helena [UNESP] Feature augmentation Image classification LSTM Manifold learning Ranking |
title_short |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
title_full |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
title_fullStr |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
title_full_unstemmed |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
title_sort |
Feature augmentation based on manifold ranking and LSTM for image classification[Formula presented] |
author |
Pereira-Ferrero, Vanessa Helena [UNESP] |
author_facet |
Pereira-Ferrero, Vanessa Helena [UNESP] Valem, Lucas Pascotti [UNESP] Pedronette, Daniel Carlos Guimarães [UNESP] |
author_role |
author |
author2 |
Valem, Lucas Pascotti [UNESP] Pedronette, Daniel Carlos Guimarães [UNESP] |
author2_role |
author author |
dc.contributor.none.fl_str_mv |
Universidade Estadual Paulista (UNESP) |
dc.contributor.author.fl_str_mv |
Pereira-Ferrero, Vanessa Helena [UNESP] Valem, Lucas Pascotti [UNESP] Pedronette, Daniel Carlos Guimarães [UNESP] |
dc.subject.por.fl_str_mv |
Feature augmentation Image classification LSTM Manifold learning Ranking |
topic |
Feature augmentation Image classification LSTM Manifold learning Ranking |
description |
Image classification is a critical topic due to its wide application and several challenges associated. Despite the huge progress made last decades, there is still a demand for context-aware image representation approaches capable of taking into the dataset manifold for improving classification accuracy. In this work, a representation learning approach is proposed, based on a novel feature augmentation strategy. The proposed method aims to exploit available contextual similarity information through rank-based manifold learning used to define and assign weights to samples used in augmentation. The approach is validated using CNN-based features and LSTM models to achieve even higher accuracy results on image classification tasks. Experimental results show that the feature augmentation strategy can indeed improve the accuracy of results on widely used image datasets (CIFAR10, Stanford Dogs, Linnaeus5, Flowers102 and Flowers17) in different CNNs (ResNet152, VGG16, DPN92). The results indicate gains up to 20% and show the potential of the developed approach in achieving higher accuracy results for image classification. |
publishDate |
2023 |
dc.date.none.fl_str_mv |
2023-07-29T13:25:50Z 2023-07-29T13:25:50Z 2023-03-01 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/article |
format |
article |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://dx.doi.org/10.1016/j.eswa.2022.118995 Expert Systems with Applications, v. 213. 0957-4174 http://hdl.handle.net/11449/247785 10.1016/j.eswa.2022.118995 2-s2.0-85140439236 |
url |
http://dx.doi.org/10.1016/j.eswa.2022.118995 http://hdl.handle.net/11449/247785 |
identifier_str_mv |
Expert Systems with Applications, v. 213. 0957-4174 10.1016/j.eswa.2022.118995 2-s2.0-85140439236 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
Expert Systems with Applications |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.source.none.fl_str_mv |
Scopus reponame:Repositório Institucional da UNESP instname:Universidade Estadual Paulista (UNESP) instacron:UNESP |
instname_str |
Universidade Estadual Paulista (UNESP) |
instacron_str |
UNESP |
institution |
UNESP |
reponame_str |
Repositório Institucional da UNESP |
collection |
Repositório Institucional da UNESP |
repository.name.fl_str_mv |
Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP) |
repository.mail.fl_str_mv |
|
_version_ |
1808128293369544704 |