Analysis of the proficiency of fully connected neural networks in the process of classifying digital images

Detalhes bibliográficos
Autor(a) principal: Janke, Jonathan
Data de Publicação: 2019
Outros Autores: Castelli, Mauro, Popovič, Aleš
Tipo de documento: Artigo
Idioma: eng
Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
Texto Completo: http://hdl.handle.net/10362/95126
Resumo: Janke, J., Castelli, M., & Popovič, A. (2019). Analysis of the proficiency of fully connected neural networks in the process of classifying digital images: Benchmark of different classification algorithms on high-level image features from convolutional layers. Expert Systems with Applications, 135, 12-38. https://doi.org/10.1016/j.eswa.2019.05.058
id RCAP_6de5979896b5ed44cd15af9310b76b25
oai_identifier_str oai:run.unl.pt:10362/95126
network_acronym_str RCAP
network_name_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository_id_str 7160
spelling Analysis of the proficiency of fully connected neural networks in the process of classifying digital imagesBenchmark of different classification algorithms on high-level image features from convolutional layersComputer visionConvolutional neural networksImage classificationEngineering(all)Computer Science ApplicationsArtificial IntelligenceJanke, J., Castelli, M., & Popovič, A. (2019). Analysis of the proficiency of fully connected neural networks in the process of classifying digital images: Benchmark of different classification algorithms on high-level image features from convolutional layers. Expert Systems with Applications, 135, 12-38. https://doi.org/10.1016/j.eswa.2019.05.058Over the course of research on convolutional neural network (CNN) architectures, few modifications have been made to the fully connected layers at the ends of the networks. In image classification, these neural network layers are responsible for creating the final classification results based on the output of the last layer of high-level image filters. Before the breakthrough of CNNs, these image filters were handcrafted, and any classification algorithm could be applied to their output. Because neural networks use gradient descent to learn their weights subject to the classification error, fully connected neural networks are a natural choice for CNNs. But a question arises: Are fully connected layers in a CNN superior to other classification algorithms? In this work, we benchmark different classification algorithms on CNNs by removing the existing fully connected classifiers. Thus, the flattened output from the last convolutional layer is used as the input for multiple benchmark classification algorithms. To ensure the generalisability of the findings, numerous CNNs are trained on CIFAR-10, CIFAR-100, and a subset of ILSVRC-2012 with 100 classes. The experimental results reveal that multiple classification algorithms, namely logistic regression, support vector machines, eXtreme gradient boosting, random forests and K-nearest neighbours, are capable of outperforming fully connected neural networks. Furthermore, the superiority of a particular classification algorithm depends on the underlying CNN structure and the nature of the classification problem. For classification problems with many classes or for CNNs that produce many high-level image features, other classification algorithms are likely to perform better than fully connected neural networks. It follows that it is advisable to benchmark multiple classification algorithms on high-level image features produced from the CNN layers to improve classification performance.NOVA Information Management School (NOVA IMS)Information Management Research Center (MagIC) - NOVA Information Management SchoolRUNJanke, JonathanCastelli, MauroPopovič, Aleš2023-03-17T01:31:48Z2019-11-302019-11-30T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/article27application/pdfhttp://hdl.handle.net/10362/95126eng0957-4174PURE: 13771460https://doi.org/10.1016/j.eswa.2019.05.058info:eu-repo/semantics/openAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2024-03-11T04:43:08Zoai:run.unl.pt:10362/95126Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-20T03:38:13.768828Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse
dc.title.none.fl_str_mv Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
Benchmark of different classification algorithms on high-level image features from convolutional layers
title Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
spellingShingle Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
Janke, Jonathan
Computer vision
Convolutional neural networks
Image classification
Engineering(all)
Computer Science Applications
Artificial Intelligence
title_short Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
title_full Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
title_fullStr Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
title_full_unstemmed Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
title_sort Analysis of the proficiency of fully connected neural networks in the process of classifying digital images
author Janke, Jonathan
author_facet Janke, Jonathan
Castelli, Mauro
Popovič, Aleš
author_role author
author2 Castelli, Mauro
Popovič, Aleš
author2_role author
author
dc.contributor.none.fl_str_mv NOVA Information Management School (NOVA IMS)
Information Management Research Center (MagIC) - NOVA Information Management School
RUN
dc.contributor.author.fl_str_mv Janke, Jonathan
Castelli, Mauro
Popovič, Aleš
dc.subject.por.fl_str_mv Computer vision
Convolutional neural networks
Image classification
Engineering(all)
Computer Science Applications
Artificial Intelligence
topic Computer vision
Convolutional neural networks
Image classification
Engineering(all)
Computer Science Applications
Artificial Intelligence
description Janke, J., Castelli, M., & Popovič, A. (2019). Analysis of the proficiency of fully connected neural networks in the process of classifying digital images: Benchmark of different classification algorithms on high-level image features from convolutional layers. Expert Systems with Applications, 135, 12-38. https://doi.org/10.1016/j.eswa.2019.05.058
publishDate 2019
dc.date.none.fl_str_mv 2019-11-30
2019-11-30T00:00:00Z
2023-03-17T01:31:48Z
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://hdl.handle.net/10362/95126
url http://hdl.handle.net/10362/95126
dc.language.iso.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv 0957-4174
PURE: 13771460
https://doi.org/10.1016/j.eswa.2019.05.058
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv 27
application/pdf
dc.source.none.fl_str_mv reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron:RCAAP
instname_str Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron_str RCAAP
institution RCAAP
reponame_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
collection Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository.name.fl_str_mv Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
repository.mail.fl_str_mv
_version_ 1799137998838693888

Registros relacionados