Random Balance ensembles for multiclass imbalance learning

Random Balance strategy (RandBal) has been recently proposed for constructing classifier ensembles for imbalanced, two-class data sets. In RandBal, each base classifier is trained with a sample of the data with a random class prevalence, independent of the a priori distribution. Hence, for each samp...

ver descrição completa

Detalhes bibliográficos
Autores: Rodríguez Diez, Juan José, Diez Pastor, José Francisco, Arnaiz González, Álvar, Kuncheva, Ludmila I. .
Formato: artículo
Estado:Versión aceptada para publicación
Fecha de publicación:2020
País:España
Recursos:Universidad de Burgos (UBU)
Repositorio:Repositorio Institucional de la Universidad de Burgos (RIUBU)
OAI Identifier:oai:riubu.ubu.es:10259/5543
Acesso em linha:http://hdl.handle.net/10259/5543
Access Level:acceso abierto
Palavra-chave:Classifier ensembles
Imbalanced data
Multiclass classification
Informática
Computer science
id ES_eecfad4d921b3a52bba548633ead4fec
oai_identifier_str oai:riubu.ubu.es:10259/5543
network_acronym_str ES
network_name_str España
repository_id_str
spelling Random Balance ensembles for multiclass imbalance learningRodríguez Diez, Juan JoséDiez Pastor, José FranciscoArnaiz González, ÁlvarKuncheva, Ludmila I. .Classifier ensemblesImbalanced dataMulticlass classificationInformáticaComputer scienceRandom Balance strategy (RandBal) has been recently proposed for constructing classifier ensembles for imbalanced, two-class data sets. In RandBal, each base classifier is trained with a sample of the data with a random class prevalence, independent of the a priori distribution. Hence, for each sample, one of the classes will be undersampled while the other will be oversampled. RandBal can be applied on its own or can be combined with any other ensemble method. One particularly successful variant is RandBalBoost which integrates Random Balance and boosting. Encouraged by the success of RandBal, this work proposes two approaches which extend RandBal to multiclass imbalance problems. Multiclass imbalance implies that at least two classes have substantially different proportion of instances. In the first approach proposed here, termed Multiple Random Balance (MultiRandBal), we deal with all classes simultaneously. The training data for each base classifier are sampled with random class proportions. The second approach we propose decomposes the multiclass problem into two-class problems using one-vs-one or one-vs-all, and builds an ensemble of RandBal ensembles. We call the two versions of the second approach OVO-RandBal and OVA-RandBal, respectively. These two approaches were chosen because they are the most straightforward extensions of RandBal for multiple classes. Our main objective is to evaluate both approaches for multiclass imbalanced problems. To this end, an experiment was carried out with 52 multiclass data sets. The results suggest that both MultiRandBal, and OVO/OVA-RandBal are viable extensions of the original two-class RandBal. Collectively, they consistently outperform acclaimed state-of-the art methods for multiclass imbalanced problems.Ministerio de Economía yCompetitividad[http://dx.doi.org/10.13039/501100003329] of theSpanishGovernmentthroughprojectTIN2015-67534-P(MINECO/FEDER, UE) and the Junta de Castilla y León through projectBU085P17 (JCyL/FEDER, UE)Elsevier202020202020info:eu-repo/semantics/articleinfo:eu-repo/semantics/acceptedVersionapplication/pdfhttp://hdl.handle.net/10259/5543reponame:Repositorio Institucional de la Universidad de Burgos (RIUBU)instname:Universidad de Burgos (UBU)InglésKnowledge-Based Systems. 2020, V. 193, 105434https://doi.org/10.1016/j.knosys.2019.105434info:eu-repo/grantAgreement/MINECO/TIN2015-67534-P/info:eu-repo/grantAgreement/JCyL/BU085P17/Attribution-NonCommercial-NoDerivatives 4.0 Internacionalhttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessoai:riubu.ubu.es:10259/55432026-05-28T07:56:11Z
dc.title.none.fl_str_mv Random Balance ensembles for multiclass imbalance learning
title Random Balance ensembles for multiclass imbalance learning
spellingShingle Random Balance ensembles for multiclass imbalance learning
Rodríguez Diez, Juan José
Classifier ensembles
Imbalanced data
Multiclass classification
Informática
Computer science
title_short Random Balance ensembles for multiclass imbalance learning
title_full Random Balance ensembles for multiclass imbalance learning
title_fullStr Random Balance ensembles for multiclass imbalance learning
title_full_unstemmed Random Balance ensembles for multiclass imbalance learning
title_sort Random Balance ensembles for multiclass imbalance learning
dc.creator.none.fl_str_mv Rodríguez Diez, Juan José
Diez Pastor, José Francisco
Arnaiz González, Álvar
Kuncheva, Ludmila I. .
author Rodríguez Diez, Juan José
author_facet Rodríguez Diez, Juan José
Diez Pastor, José Francisco
Arnaiz González, Álvar
Kuncheva, Ludmila I. .
author_role author
author2 Diez Pastor, José Francisco
Arnaiz González, Álvar
Kuncheva, Ludmila I. .
author2_role author
author
author
dc.subject.none.fl_str_mv Classifier ensembles
Imbalanced data
Multiclass classification
Informática
Computer science
topic Classifier ensembles
Imbalanced data
Multiclass classification
Informática
Computer science
description Random Balance strategy (RandBal) has been recently proposed for constructing classifier ensembles for imbalanced, two-class data sets. In RandBal, each base classifier is trained with a sample of the data with a random class prevalence, independent of the a priori distribution. Hence, for each sample, one of the classes will be undersampled while the other will be oversampled. RandBal can be applied on its own or can be combined with any other ensemble method. One particularly successful variant is RandBalBoost which integrates Random Balance and boosting. Encouraged by the success of RandBal, this work proposes two approaches which extend RandBal to multiclass imbalance problems. Multiclass imbalance implies that at least two classes have substantially different proportion of instances. In the first approach proposed here, termed Multiple Random Balance (MultiRandBal), we deal with all classes simultaneously. The training data for each base classifier are sampled with random class proportions. The second approach we propose decomposes the multiclass problem into two-class problems using one-vs-one or one-vs-all, and builds an ensemble of RandBal ensembles. We call the two versions of the second approach OVO-RandBal and OVA-RandBal, respectively. These two approaches were chosen because they are the most straightforward extensions of RandBal for multiple classes. Our main objective is to evaluate both approaches for multiclass imbalanced problems. To this end, an experiment was carried out with 52 multiclass data sets. The results suggest that both MultiRandBal, and OVO/OVA-RandBal are viable extensions of the original two-class RandBal. Collectively, they consistently outperform acclaimed state-of-the art methods for multiclass imbalanced problems.
publishDate 2020
dc.date.none.fl_str_mv 2020
2020
2020
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/acceptedVersion
format article
status_str acceptedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/10259/5543
url http://hdl.handle.net/10259/5543
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv Knowledge-Based Systems. 2020, V. 193, 105434
https://doi.org/10.1016/j.knosys.2019.105434
info:eu-repo/grantAgreement/MINECO/TIN2015-67534-P/
info:eu-repo/grantAgreement/JCyL/BU085P17/
dc.rights.none.fl_str_mv Attribution-NonCommercial-NoDerivatives 4.0 Internacional
http://creativecommons.org/licenses/by-nc-nd/4.0/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv Attribution-NonCommercial-NoDerivatives 4.0 Internacional
http://creativecommons.org/licenses/by-nc-nd/4.0/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.publisher.none.fl_str_mv Elsevier
publisher.none.fl_str_mv Elsevier
dc.source.none.fl_str_mv reponame:Repositorio Institucional de la Universidad de Burgos (RIUBU)
instname:Universidad de Burgos (UBU)
instname_str Universidad de Burgos (UBU)
reponame_str Repositorio Institucional de la Universidad de Burgos (RIUBU)
collection Repositorio Institucional de la Universidad de Burgos (RIUBU)
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869423780469669889
score 15,301629

Registos relacionados