Applying artificial neural networks as a test to detect milk fraud by whey addition

Detalhes bibliográficos
Autor(a) principal: Valente, Gerson de Freitas Silva
Data de Publicação: 2014
Outros Autores: Guimarães, Daiana Cristina, Gaspardi, Ana Laís Andrade, Oliveira, Lara de Andrade
Tipo de documento: Artigo
Idioma: por
Título da fonte: Revista do Instituto de Laticínios Cândido Tostes
Texto Completo: https://www.revistadoilct.com.br/rilct/article/view/353
Resumo: This study aimed to employ Artificial Neural Networks to classify milk samples from routine analysis of a dairy company in order to identify adulteration by whey addition. The samples were prepared by mixing the milk with different whey concentrations (0, 1, 5, 10, and 20%), which were then analyzed for temperature, fat content, solids-non-fat, bulk density, protein, lactose, minerals, freezing point, conductivity, and pH, for a total of 167 assays. Out of these, 101 were used to train the network, 33 for validation, and 33 to test the artificial neural network. The best classification was obtained using a radial basis function neural network. k-means algorithm was used to obtain the network center, k-nearest was used to define the receptive fields, and the pseudo-inverse method was used to define the weights of the output layer. The best result was found with a network with 10 neurons in the input layer, 40 neurons in the hidden layer, and two neurons in the output layer, achieving over 95% accuracy in classification. The classification methodology using artificial neural networks has strong potential to be applied in interpreting data from routine analysis in dairy companies in order to classify milk adulterated with whey and, later, confirm the result using official methodologies.
id EPAMIG-1_d63e2b7442f1525aa24bbe9c5f64d270
oai_identifier_str oai:oai.rilct.emnuvens.com.br:article/353
network_acronym_str EPAMIG-1
network_name_str Revista do Instituto de Laticínios Cândido Tostes
repository_id_str
spelling Applying artificial neural networks as a test to detect milk fraud by whey additionAPLICAÇÃO DE REDES NEURAIS ARTIFICIAIS COMO TESTE DE DETECÇÃO DE FRAUDE DE LEITE POR ADIÇÃO DE SORO DE QUEIJOclassification; physicochemical analysis; RBFControle de Qualidade; Engenharia.classificação; análise físico-química; RBF.This study aimed to employ Artificial Neural Networks to classify milk samples from routine analysis of a dairy company in order to identify adulteration by whey addition. The samples were prepared by mixing the milk with different whey concentrations (0, 1, 5, 10, and 20%), which were then analyzed for temperature, fat content, solids-non-fat, bulk density, protein, lactose, minerals, freezing point, conductivity, and pH, for a total of 167 assays. Out of these, 101 were used to train the network, 33 for validation, and 33 to test the artificial neural network. The best classification was obtained using a radial basis function neural network. k-means algorithm was used to obtain the network center, k-nearest was used to define the receptive fields, and the pseudo-inverse method was used to define the weights of the output layer. The best result was found with a network with 10 neurons in the input layer, 40 neurons in the hidden layer, and two neurons in the output layer, achieving over 95% accuracy in classification. The classification methodology using artificial neural networks has strong potential to be applied in interpreting data from routine analysis in dairy companies in order to classify milk adulterated with whey and, later, confirm the result using official methodologies.Esse trabalho foi realizado com o objetivo de empregar Redes Neurais Artificiais para classificar amostras de leite, a partir de análises de rotina de um laticínio, em amostras de leite adulteradas ou não, quanto à adição de soro de queijo. As amostras foram preparadas através da mistura do leite com diferentes concentrações de soro (0, 1, 5, 10 e 20%) sendo analisadas quanto a temperatura, teor de gordura, extrato seco desengordurado, densidade, proteína, lactose, sais minerais, ponto de congelamento, condutividade e pH, totalizando 167 ensaios. Desses 167 ensaios, 101 foram usados para treinamento da rede, 33 para validação e outros 33 para testar a rede neural artificial. A melhor rede de classificação foi uma rede neural de função de base radial. Para obter os centros da rede foi usado o algoritmo k-means, para definir a largura dos campos receptivos o k-nearest e os pesos da camada de saída foram definidos usando o método da pseudo-inversa. A rede com melhor resultado apresentou 10 neurônios na camada de entrada, 40 neurônios na camada oculta e dois na camada de saída, sendo possível obter mais de 95% de acertos na classificação. A metodologia de classificação por Redes Neurais Artificiais apresenta um grande potencial de aplicação na interpretação de dados de análises de rotina nos laticínios para classificação do leite em adulterado por soro de queijo e, posteriormente, confirmação do resultado por metodologia oficial.ILCTIFSULDEMINAS-Câmpus InconfidentesCNPqValente, Gerson de Freitas SilvaGuimarães, Daiana CristinaGaspardi, Ana Laís AndradeOliveira, Lara de Andrade2014-12-02info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfhttps://www.revistadoilct.com.br/rilct/article/view/35310.14295/2238-6416.v69i6.353Journal of Candido Tostes Dairy Institute; v. 69, n. 6 (2014); 425-432Revista do Instituto de Laticínios Cândido Tostes; v. 69, n. 6 (2014); 425-4322238-64160100-3674reponame:Revista do Instituto de Laticínios Cândido Tostesinstname:Empresa de Pesquisa Agropecuária de Minas Gerais (EPAMIG)instacron:EPAMIGporhttps://www.revistadoilct.com.br/rilct/article/view/353/344Direitos autorais 2015 Revista do Instituto de Laticínios Cândido Tosteshttp://creativecommons.org/licenses/by/4.0info:eu-repo/semantics/openAccess2015-03-09T18:09:06Zoai:oai.rilct.emnuvens.com.br:article/353Revistahttp://www.revistadoilct.com.br/ONGhttps://www.revistadoilct.com.br/rilct/oai||revistadoilct@epamig.br|| revistadoilct@oi.com.br2238-64160100-3674opendoar:2015-03-09T18:09:06Revista do Instituto de Laticínios Cândido Tostes - Empresa de Pesquisa Agropecuária de Minas Gerais (EPAMIG)false
dc.title.none.fl_str_mv Applying artificial neural networks as a test to detect milk fraud by whey addition
APLICAÇÃO DE REDES NEURAIS ARTIFICIAIS COMO TESTE DE DETECÇÃO DE FRAUDE DE LEITE POR ADIÇÃO DE SORO DE QUEIJO
title Applying artificial neural networks as a test to detect milk fraud by whey addition
spellingShingle Applying artificial neural networks as a test to detect milk fraud by whey addition
Valente, Gerson de Freitas Silva
classification; physicochemical analysis; RBF
Controle de Qualidade; Engenharia.
classificação; análise físico-química; RBF.
title_short Applying artificial neural networks as a test to detect milk fraud by whey addition
title_full Applying artificial neural networks as a test to detect milk fraud by whey addition
title_fullStr Applying artificial neural networks as a test to detect milk fraud by whey addition
title_full_unstemmed Applying artificial neural networks as a test to detect milk fraud by whey addition
title_sort Applying artificial neural networks as a test to detect milk fraud by whey addition
author Valente, Gerson de Freitas Silva
author_facet Valente, Gerson de Freitas Silva
Guimarães, Daiana Cristina
Gaspardi, Ana Laís Andrade
Oliveira, Lara de Andrade
author_role author
author2 Guimarães, Daiana Cristina
Gaspardi, Ana Laís Andrade
Oliveira, Lara de Andrade
author2_role author
author
author
dc.contributor.none.fl_str_mv
IFSULDEMINAS-Câmpus Inconfidentes
CNPq
dc.contributor.author.fl_str_mv Valente, Gerson de Freitas Silva
Guimarães, Daiana Cristina
Gaspardi, Ana Laís Andrade
Oliveira, Lara de Andrade
dc.subject.none.fl_str_mv
dc.subject.por.fl_str_mv classification; physicochemical analysis; RBF
Controle de Qualidade; Engenharia.
classificação; análise físico-química; RBF.
topic classification; physicochemical analysis; RBF
Controle de Qualidade; Engenharia.
classificação; análise físico-química; RBF.
description This study aimed to employ Artificial Neural Networks to classify milk samples from routine analysis of a dairy company in order to identify adulteration by whey addition. The samples were prepared by mixing the milk with different whey concentrations (0, 1, 5, 10, and 20%), which were then analyzed for temperature, fat content, solids-non-fat, bulk density, protein, lactose, minerals, freezing point, conductivity, and pH, for a total of 167 assays. Out of these, 101 were used to train the network, 33 for validation, and 33 to test the artificial neural network. The best classification was obtained using a radial basis function neural network. k-means algorithm was used to obtain the network center, k-nearest was used to define the receptive fields, and the pseudo-inverse method was used to define the weights of the output layer. The best result was found with a network with 10 neurons in the input layer, 40 neurons in the hidden layer, and two neurons in the output layer, achieving over 95% accuracy in classification. The classification methodology using artificial neural networks has strong potential to be applied in interpreting data from routine analysis in dairy companies in order to classify milk adulterated with whey and, later, confirm the result using official methodologies.
publishDate 2014
dc.date.none.fl_str_mv 2014-12-02
dc.type.none.fl_str_mv


dc.type.driver.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.uri.fl_str_mv https://www.revistadoilct.com.br/rilct/article/view/353
10.14295/2238-6416.v69i6.353
url https://www.revistadoilct.com.br/rilct/article/view/353
identifier_str_mv 10.14295/2238-6416.v69i6.353
dc.language.iso.fl_str_mv por
language por
dc.relation.none.fl_str_mv https://www.revistadoilct.com.br/rilct/article/view/353/344
dc.rights.driver.fl_str_mv Direitos autorais 2015 Revista do Instituto de Laticínios Cândido Tostes
http://creativecommons.org/licenses/by/4.0
info:eu-repo/semantics/openAccess
rights_invalid_str_mv Direitos autorais 2015 Revista do Instituto de Laticínios Cândido Tostes
http://creativecommons.org/licenses/by/4.0
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.publisher.none.fl_str_mv ILCT
publisher.none.fl_str_mv ILCT
dc.source.none.fl_str_mv Journal of Candido Tostes Dairy Institute; v. 69, n. 6 (2014); 425-432
Revista do Instituto de Laticínios Cândido Tostes; v. 69, n. 6 (2014); 425-432
2238-6416
0100-3674
reponame:Revista do Instituto de Laticínios Cândido Tostes
instname:Empresa de Pesquisa Agropecuária de Minas Gerais (EPAMIG)
instacron:EPAMIG
instname_str Empresa de Pesquisa Agropecuária de Minas Gerais (EPAMIG)
instacron_str EPAMIG
institution EPAMIG
reponame_str Revista do Instituto de Laticínios Cândido Tostes
collection Revista do Instituto de Laticínios Cândido Tostes
repository.name.fl_str_mv Revista do Instituto de Laticínios Cândido Tostes - Empresa de Pesquisa Agropecuária de Minas Gerais (EPAMIG)
repository.mail.fl_str_mv ||revistadoilct@epamig.br|| revistadoilct@oi.com.br
_version_ 1809738130911133696