Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements
Ano de defesa: | 2021 |
---|---|
Autor(a) principal: | |
Orientador(a): | |
Banca de defesa: | |
Tipo de documento: | Tese |
Tipo de acesso: | Acesso aberto |
Idioma: | eng |
Instituição de defesa: |
Universidade Federal de Itajubá
|
Programa de Pós-Graduação: |
Programa de Pós-Graduação: Doutorado - Engenharia Elétrica
|
Departamento: |
IESTI - Instituto de Engenharia de Sistemas e Tecnologia da Informação
|
País: |
Brasil
|
Palavras-chave em Português: | |
Área do conhecimento CNPq: | |
Link de acesso: | https://repositorio.unifei.edu.br/jspui/handle/123456789/2477 |
Resumo: | Pattern recognition techniques in the Myoelectric Signal (EMG) are employed in the development of robotic prostheses, and for that, they adopt several approaches of Artificial Intelligence (AI). This Thesis proposes to solve the problem of recognition of EMG standards through the adoption of profound learning techniques in an optimized way. The research developed an approach that extracts the characteristic a priori to feed the classifiers that supposedly do not need this step. The study integrated the BioPatRec platform (advanced prosthesis study and development) to two classification algorithms (Convolutional Neural Network and Long Short-Term Memory) in a hybrid way, where the input provided to the network already has characteristics that describe the movement (level of muscle activation, magnitude, amplitude, power, and others). Thus, the signal is tracked as a time series instead of an image, which allows us to eliminate a set of points irrelevant to the classifier, making the information expressive. In the sequence, the methodology developed software that implements the concept introduced using a Graphical Processing Unit (GPU) in parallel this increment allowed the classification model to combine high precision with a training time of less than 1 second. The parallel model was called BioPatRec-Py and employed some Engineering techniques of Features that managed to make the network entry more homogeneous, reducing variability, noise, and standardizing distribution. The research obtained satisfactory results and surpassed the other classification algorithms in most of the evaluated experiments. The work performed a statistical analysis of the outcomes and fine-tuned the hyperparameters of each of the networks. Ultimately, BioPatRec-Py provided a generic model. The network was trained globally between individuals, allowing the creation of a standardized approach, with an average accuracy of 97.83%. |
id |
UFEI_87e3884715b0768fa8955e9c914f27a3 |
---|---|
oai_identifier_str |
oai:repositorio.unifei.edu.br:123456789/2477 |
network_acronym_str |
UFEI |
network_name_str |
Repositório Institucional da UNIFEI (RIUNIFEI) |
repository_id_str |
|
spelling |
2021-07-132021-08-022021-08-02T21:36:48Z2021-08-02T21:36:48Zhttps://repositorio.unifei.edu.br/jspui/handle/123456789/2477Pattern recognition techniques in the Myoelectric Signal (EMG) are employed in the development of robotic prostheses, and for that, they adopt several approaches of Artificial Intelligence (AI). This Thesis proposes to solve the problem of recognition of EMG standards through the adoption of profound learning techniques in an optimized way. The research developed an approach that extracts the characteristic a priori to feed the classifiers that supposedly do not need this step. The study integrated the BioPatRec platform (advanced prosthesis study and development) to two classification algorithms (Convolutional Neural Network and Long Short-Term Memory) in a hybrid way, where the input provided to the network already has characteristics that describe the movement (level of muscle activation, magnitude, amplitude, power, and others). Thus, the signal is tracked as a time series instead of an image, which allows us to eliminate a set of points irrelevant to the classifier, making the information expressive. In the sequence, the methodology developed software that implements the concept introduced using a Graphical Processing Unit (GPU) in parallel this increment allowed the classification model to combine high precision with a training time of less than 1 second. The parallel model was called BioPatRec-Py and employed some Engineering techniques of Features that managed to make the network entry more homogeneous, reducing variability, noise, and standardizing distribution. The research obtained satisfactory results and surpassed the other classification algorithms in most of the evaluated experiments. The work performed a statistical analysis of the outcomes and fine-tuned the hyperparameters of each of the networks. Ultimately, BioPatRec-Py provided a generic model. The network was trained globally between individuals, allowing the creation of a standardized approach, with an average accuracy of 97.83%.Técnicas de reconhecimento de padrões no Sinal Mioelétrico (EMG) são empregadas no desenvolvimento de próteses robóticas, e para isso, adotam diversas abordagens de Inteligência Artificial (IA). Esta Tese se propõe a resolver o problema de reconhecimento de padrões EMG através da adoção de técnicas de aprendizado profundo de forma otimizada. Para isso, desenvolveu uma abordagem que realiza a extração da característica a priori, para alimentar os classificadores que supostamente não necessitam dessa etapa. O estudo integrou a plataforma BioPatRec (estudo e desenvolvimento avançado de próteses) a dois algoritmos de classificação (Convolutional Neural Network e Long Short-Term Memory) de forma híbrida, onde a entrada fornecida à rede já possui características que descrevem o movimento (nível de ativação muscular, magnitude, amplitude, potência e outros). Assim, o sinal é rastreado como uma série temporal ao invés de uma imagem, o que nos permite eliminar um conjunto de pontos irrelevantes para o classificador, tornando a informação expressivas. Na sequência, a metodologia desenvolveu um software que implementa o conceito introduzido utilizando uma Unidade de Processamento Gráfico (GPU) de modo paralelo, esse incremento permitiu que o modelo de classificação aliasse alta precisão com um tempo de treinamento inferior a 1 segundo. O modelo paralelizado foi chamado de BioPatRec-Py e empregou algumas técnicas de Engenharia de Features que conseguiram tornar a entrada da rede mais homogênea, reduzindo a variabilidade, o ruído e uniformizando a distribuição. A pesquisa obteve resultados satisfatórios e superou os demais algoritmos de classificação na maioria dos experimentos avaliados. O trabalho também realizou uma análise estatística dos resultados e fez o ajuste fino dos hiper-parâmetros de cada uma das redes. Em última instancia, o BioPatRec-Py forneceu um modelo genérico. A rede foi treinada globalmente entre os indivíduos, permitindo a criação de uma abordagem global, com uma precisão média de 97,83%.engUniversidade Federal de ItajubáPrograma de Pós-Graduação: Doutorado - Engenharia ElétricaUNIFEIBrasilIESTI - Instituto de Engenharia de Sistemas e Tecnologia da InformaçãoCNPQ::ENGENHARIAS::ENGENHARIA ELÉTRICABio-sinaisBioPatRecCNNEngenharia de featureEngenharia de reabilitaçãoLSTMDeep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movementsinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/doctoralThesisMORENO, Robson Luizhttp://lattes.cnpq.br/6281644588548940PIMENTA, Tales Cleberhttp://lattes.cnpq.br/3321577431881283http://lattes.cnpq.br/9282621947516871SOUZA, Gabriel Cirac MendesSOUZA, Gabriel Cirac Mendes. Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements. 2021. 145 f. Tese (Doutorado em Engenharia Elétrica) – Universidade Federal de Itajubá, Itajubá, 2021.info:eu-repo/semantics/openAccessreponame:Repositório Institucional da UNIFEI (RIUNIFEI)instname:Universidade Federal de Itajubá (UNIFEI)instacron:UNIFEILICENSElicense.txtlicense.txttext/plain; charset=utf-81748https://repositorio.unifei.edu.br/jspui/bitstream/123456789/2477/2/license.txt8a4605be74aa9ea9d79846c1fba20a33MD52ORIGINALTese_2021037.pdfTese_2021037.pdfapplication/pdf5942977https://repositorio.unifei.edu.br/jspui/bitstream/123456789/2477/1/Tese_2021037.pdf91678405d0554f9be5f585666e0c9ff3MD51123456789/24772021-08-09 11:32:38.984oai:repositorio.unifei.edu.br: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Repositório InstitucionalPUBhttps://repositorio.unifei.edu.br/oai/requestrepositorio@unifei.edu.br || geraldocarlos@unifei.edu.bropendoar:70442021-08-09T14:32:38Repositório Institucional da UNIFEI (RIUNIFEI) - Universidade Federal de Itajubá (UNIFEI)false |
dc.title.pt_BR.fl_str_mv |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
title |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
spellingShingle |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements SOUZA, Gabriel Cirac Mendes CNPQ::ENGENHARIAS::ENGENHARIA ELÉTRICA Bio-sinais BioPatRec CNN Engenharia de feature Engenharia de reabilitação LSTM |
title_short |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
title_full |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
title_fullStr |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
title_full_unstemmed |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
title_sort |
Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements |
author |
SOUZA, Gabriel Cirac Mendes |
author_facet |
SOUZA, Gabriel Cirac Mendes |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
MORENO, Robson Luiz |
dc.contributor.advisor1Lattes.fl_str_mv |
http://lattes.cnpq.br/6281644588548940 |
dc.contributor.advisor-co1.fl_str_mv |
PIMENTA, Tales Cleber |
dc.contributor.advisor-co1Lattes.fl_str_mv |
http://lattes.cnpq.br/3321577431881283 |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/9282621947516871 |
dc.contributor.author.fl_str_mv |
SOUZA, Gabriel Cirac Mendes |
contributor_str_mv |
MORENO, Robson Luiz PIMENTA, Tales Cleber |
dc.subject.cnpq.fl_str_mv |
CNPQ::ENGENHARIAS::ENGENHARIA ELÉTRICA |
topic |
CNPQ::ENGENHARIAS::ENGENHARIA ELÉTRICA Bio-sinais BioPatRec CNN Engenharia de feature Engenharia de reabilitação LSTM |
dc.subject.por.fl_str_mv |
Bio-sinais BioPatRec CNN Engenharia de feature Engenharia de reabilitação LSTM |
description |
Pattern recognition techniques in the Myoelectric Signal (EMG) are employed in the development of robotic prostheses, and for that, they adopt several approaches of Artificial Intelligence (AI). This Thesis proposes to solve the problem of recognition of EMG standards through the adoption of profound learning techniques in an optimized way. The research developed an approach that extracts the characteristic a priori to feed the classifiers that supposedly do not need this step. The study integrated the BioPatRec platform (advanced prosthesis study and development) to two classification algorithms (Convolutional Neural Network and Long Short-Term Memory) in a hybrid way, where the input provided to the network already has characteristics that describe the movement (level of muscle activation, magnitude, amplitude, power, and others). Thus, the signal is tracked as a time series instead of an image, which allows us to eliminate a set of points irrelevant to the classifier, making the information expressive. In the sequence, the methodology developed software that implements the concept introduced using a Graphical Processing Unit (GPU) in parallel this increment allowed the classification model to combine high precision with a training time of less than 1 second. The parallel model was called BioPatRec-Py and employed some Engineering techniques of Features that managed to make the network entry more homogeneous, reducing variability, noise, and standardizing distribution. The research obtained satisfactory results and surpassed the other classification algorithms in most of the evaluated experiments. The work performed a statistical analysis of the outcomes and fine-tuned the hyperparameters of each of the networks. Ultimately, BioPatRec-Py provided a generic model. The network was trained globally between individuals, allowing the creation of a standardized approach, with an average accuracy of 97.83%. |
publishDate |
2021 |
dc.date.issued.fl_str_mv |
2021-07-13 |
dc.date.available.fl_str_mv |
2021-08-02 2021-08-02T21:36:48Z |
dc.date.accessioned.fl_str_mv |
2021-08-02T21:36:48Z |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/doctoralThesis |
format |
doctoralThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
https://repositorio.unifei.edu.br/jspui/handle/123456789/2477 |
url |
https://repositorio.unifei.edu.br/jspui/handle/123456789/2477 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.references.pt_BR.fl_str_mv |
SOUZA, Gabriel Cirac Mendes. Deep learning and feature engineering techniques applied to the myoelectric signal for accurate prediction of movements. 2021. 145 f. Tese (Doutorado em Engenharia Elétrica) – Universidade Federal de Itajubá, Itajubá, 2021. |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.publisher.none.fl_str_mv |
Universidade Federal de Itajubá |
dc.publisher.program.fl_str_mv |
Programa de Pós-Graduação: Doutorado - Engenharia Elétrica |
dc.publisher.initials.fl_str_mv |
UNIFEI |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
IESTI - Instituto de Engenharia de Sistemas e Tecnologia da Informação |
publisher.none.fl_str_mv |
Universidade Federal de Itajubá |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UNIFEI (RIUNIFEI) instname:Universidade Federal de Itajubá (UNIFEI) instacron:UNIFEI |
instname_str |
Universidade Federal de Itajubá (UNIFEI) |
instacron_str |
UNIFEI |
institution |
UNIFEI |
reponame_str |
Repositório Institucional da UNIFEI (RIUNIFEI) |
collection |
Repositório Institucional da UNIFEI (RIUNIFEI) |
bitstream.url.fl_str_mv |
https://repositorio.unifei.edu.br/jspui/bitstream/123456789/2477/2/license.txt https://repositorio.unifei.edu.br/jspui/bitstream/123456789/2477/1/Tese_2021037.pdf |
bitstream.checksum.fl_str_mv |
8a4605be74aa9ea9d79846c1fba20a33 91678405d0554f9be5f585666e0c9ff3 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UNIFEI (RIUNIFEI) - Universidade Federal de Itajubá (UNIFEI) |
repository.mail.fl_str_mv |
repositorio@unifei.edu.br || geraldocarlos@unifei.edu.br |
_version_ |
1797311987032522752 |