Oversampling the Minority Class in the Feature Space

The imbalanced nature of some real-world data is one of the current challenges for machine learning researchers. One common approach oversamples the minority class through convex combination of its patterns. We explore the general idea of synthetic oversampling in the feature space induced by a kern...

Descripción completa

Detalles Bibliográficos
Autores: Pérez Ortiz, María, Gutiérrez Peña, Pedro Antonio, Tino, Peter, Hervás Martínez, César
Tipo de recurso: artículo
Fecha de publicación:2016
País:España
Institución:Universidad Loyola Andalucía
Repositorio:Brújula
OAI Identifier:oai:repositorio.uloyola.es:20.500.12412/1223
Acceso en línea:http://hdl.handle.net/20.500.12412/1223
Access Level:acceso abierto
Palabra clave:Empirical feature space (EFS)
Imbalanced classification
Kernel methods
Oversampling
Support vector machines (SVMs)
id ES_f5f71184b8a7e2f9ad23f7e99cbc8ed9
oai_identifier_str oai:repositorio.uloyola.es:20.500.12412/1223
network_acronym_str ES
network_name_str España
repository_id_str
spelling Oversampling the Minority Class in the Feature SpacePérez Ortiz, MaríaGutiérrez Peña, Pedro AntonioTino, PeterHervás Martínez, CésarEmpirical feature space (EFS)Imbalanced classificationKernel methodsOversamplingSupport vector machines (SVMs)The imbalanced nature of some real-world data is one of the current challenges for machine learning researchers. One common approach oversamples the minority class through convex combination of its patterns. We explore the general idea of synthetic oversampling in the feature space induced by a kernel function (as opposed to input space). If the kernel function matches the underlying problem, the classes will be linearly separable and synthetically generated patterns will lie on the minority class region. Since the feature space is not directly accessible, we use the empirical feature space (EFS) (a Euclidean space isomorphic to the feature space) for oversampling purposes. The proposed method is framed in the context of support vector machines, where the imbalanced data sets can pose a serious hindrance. The idea is investigated in three scenarios: 1) oversampling in the full and reduced-rank EFSs; 2) a kernel learning technique maximizing the data class separation to study the influence of the feature space structure (implicitly defined by the kernel function); and 3) a unified framework for preferential oversampling that spans some of the previous approaches in the literature. We support our investigation with extensive experiments over 50 imbalanced data sets.2016info:eu-repo/semantics/articlehttp://hdl.handle.net/20.500.12412/1223reponame:Brújulainstname:Universidad Loyola AndalucíaIngléshttp://creativecommons.org/licenses/by-nc-nd/4.0/info:eu-repo/semantics/openAccessoai:repositorio.uloyola.es:20.500.12412/12232026-06-24T12:48:37Z
dc.title.none.fl_str_mv Oversampling the Minority Class in the Feature Space
title Oversampling the Minority Class in the Feature Space
spellingShingle Oversampling the Minority Class in the Feature Space
Pérez Ortiz, María
Empirical feature space (EFS)
Imbalanced classification
Kernel methods
Oversampling
Support vector machines (SVMs)
title_short Oversampling the Minority Class in the Feature Space
title_full Oversampling the Minority Class in the Feature Space
title_fullStr Oversampling the Minority Class in the Feature Space
title_full_unstemmed Oversampling the Minority Class in the Feature Space
title_sort Oversampling the Minority Class in the Feature Space
dc.creator.none.fl_str_mv Pérez Ortiz, María
Gutiérrez Peña, Pedro Antonio
Tino, Peter
Hervás Martínez, César
author Pérez Ortiz, María
author_facet Pérez Ortiz, María
Gutiérrez Peña, Pedro Antonio
Tino, Peter
Hervás Martínez, César
author_role author
author2 Gutiérrez Peña, Pedro Antonio
Tino, Peter
Hervás Martínez, César
author2_role author
author
author
dc.subject.none.fl_str_mv Empirical feature space (EFS)
Imbalanced classification
Kernel methods
Oversampling
Support vector machines (SVMs)
topic Empirical feature space (EFS)
Imbalanced classification
Kernel methods
Oversampling
Support vector machines (SVMs)
description The imbalanced nature of some real-world data is one of the current challenges for machine learning researchers. One common approach oversamples the minority class through convex combination of its patterns. We explore the general idea of synthetic oversampling in the feature space induced by a kernel function (as opposed to input space). If the kernel function matches the underlying problem, the classes will be linearly separable and synthetically generated patterns will lie on the minority class region. Since the feature space is not directly accessible, we use the empirical feature space (EFS) (a Euclidean space isomorphic to the feature space) for oversampling purposes. The proposed method is framed in the context of support vector machines, where the imbalanced data sets can pose a serious hindrance. The idea is investigated in three scenarios: 1) oversampling in the full and reduced-rank EFSs; 2) a kernel learning technique maximizing the data class separation to study the influence of the feature space structure (implicitly defined by the kernel function); and 3) a unified framework for preferential oversampling that spans some of the previous approaches in the literature. We support our investigation with extensive experiments over 50 imbalanced data sets.
publishDate 2016
dc.date.none.fl_str_mv 2016
dc.type.none.fl_str_mv info:eu-repo/semantics/article
format article
dc.identifier.none.fl_str_mv http://hdl.handle.net/20.500.12412/1223
url http://hdl.handle.net/20.500.12412/1223
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.rights.none.fl_str_mv http://creativecommons.org/licenses/by-nc-nd/4.0/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv http://creativecommons.org/licenses/by-nc-nd/4.0/
eu_rights_str_mv openAccess
dc.source.none.fl_str_mv reponame:Brújula
instname:Universidad Loyola Andalucía
instname_str Universidad Loyola Andalucía
reponame_str Brújula
collection Brújula
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869424665593643008
score 15.812429

Similares en LA Referencia