Optimal threshold estimation for binary classifiers using game theory

Autores
Sánchez Miguel, Ignacio Enrique
Año de publicación
2016
Idioma
inglés
Tipo de recurso
artículo
Estado
versión publicada
Descripción
Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs.
Fil: Sánchez Miguel, Ignacio Enrique. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales; Argentina
Materia
GAME THEORY
ROC CURVE
MINIMAX PRINCIPLE
BIONFORMATICS
Nivel de accesibilidad
acceso abierto
Condiciones de uso
https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
Repositorio
CONICET Digital (CONICET)
Institución
Consejo Nacional de Investigaciones Científicas y Técnicas
OAI Identificador
oai:ri.conicet.gov.ar:11336/60009

id CONICETDig_28982e4f9ab685ba00eab8da06c34409
oai_identifier_str oai:ri.conicet.gov.ar:11336/60009
network_acronym_str CONICETDig
repository_id_str 3498
network_name_str CONICET Digital (CONICET)
spelling Optimal threshold estimation for binary classifiers using game theorySánchez Miguel, Ignacio EnriqueGAME THEORYROC CURVEMINIMAX PRINCIPLEBIONFORMATICShttps://purl.org/becyt/ford/1.6https://purl.org/becyt/ford/1Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs.Fil: Sánchez Miguel, Ignacio Enrique. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales; ArgentinaF1000 Research Ltd2016-11-25info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionhttp://purl.org/coar/resource_type/c_6501info:ar-repo/semantics/articuloapplication/pdfapplication/pdfapplication/pdfhttp://hdl.handle.net/11336/60009Sánchez Miguel, Ignacio Enrique; Optimal threshold estimation for binary classifiers using game theory; F1000 Research Ltd; F1000Research; 5; 2762; 25-11-2016; 1-122046-1402CONICET DigitalCONICETenginfo:eu-repo/semantics/altIdentifier/url/https://f1000research.com/articles/5-2762/v3info:eu-repo/semantics/altIdentifier/doi/10.12688/f1000research.10114.3info:eu-repo/semantics/openAccesshttps://creativecommons.org/licenses/by-nc-sa/2.5/ar/reponame:CONICET Digital (CONICET)instname:Consejo Nacional de Investigaciones Científicas y Técnicas2025-09-29T10:08:31Zoai:ri.conicet.gov.ar:11336/60009instacron:CONICETInstitucionalhttp://ri.conicet.gov.ar/Organismo científico-tecnológicoNo correspondehttp://ri.conicet.gov.ar/oai/requestdasensio@conicet.gov.ar; lcarlino@conicet.gov.arArgentinaNo correspondeNo correspondeNo correspondeopendoar:34982025-09-29 10:08:32.263CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicasfalse
dc.title.none.fl_str_mv Optimal threshold estimation for binary classifiers using game theory
title Optimal threshold estimation for binary classifiers using game theory
spellingShingle Optimal threshold estimation for binary classifiers using game theory
Sánchez Miguel, Ignacio Enrique
GAME THEORY
ROC CURVE
MINIMAX PRINCIPLE
BIONFORMATICS
title_short Optimal threshold estimation for binary classifiers using game theory
title_full Optimal threshold estimation for binary classifiers using game theory
title_fullStr Optimal threshold estimation for binary classifiers using game theory
title_full_unstemmed Optimal threshold estimation for binary classifiers using game theory
title_sort Optimal threshold estimation for binary classifiers using game theory
dc.creator.none.fl_str_mv Sánchez Miguel, Ignacio Enrique
author Sánchez Miguel, Ignacio Enrique
author_facet Sánchez Miguel, Ignacio Enrique
author_role author
dc.subject.none.fl_str_mv GAME THEORY
ROC CURVE
MINIMAX PRINCIPLE
BIONFORMATICS
topic GAME THEORY
ROC CURVE
MINIMAX PRINCIPLE
BIONFORMATICS
purl_subject.fl_str_mv https://purl.org/becyt/ford/1.6
https://purl.org/becyt/ford/1
dc.description.none.fl_txt_mv Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs.
Fil: Sánchez Miguel, Ignacio Enrique. Consejo Nacional de Investigaciones Científicas y Técnicas. Oficina de Coordinación Administrativa Ciudad Universitaria. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales. Universidad de Buenos Aires. Facultad de Ciencias Exactas y Naturales. Instituto de Química Biológica de la Facultad de Ciencias Exactas y Naturales; Argentina
description Many bioinformatics algorithms can be understood as binary classifiers. They are usually compared using the area under the receiver operating characteristic (ROC) curve. On the other hand, choosing the best threshold for practical use is a complex task, due to uncertain and context-dependent skews in the abundance of positives in nature and in the yields/costs for correct/incorrect classification. We argue that considering a classifier as a player in a zero-sum game allows us to use the minimax principle from game theory to determine the optimal operating point. The proposed classifier threshold corresponds to the intersection between the ROC curve and the descending diagonal in ROC space and yields a minimax accuracy of 1-FPR. Our proposal can be readily implemented in practice, and reveals that the empirical condition for threshold estimation of “specificity equals sensitivity” maximizes robustness against uncertainties in the abundance of positives in nature and classification costs.
publishDate 2016
dc.date.none.fl_str_mv 2016-11-25
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
http://purl.org/coar/resource_type/c_6501
info:ar-repo/semantics/articulo
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/11336/60009
Sánchez Miguel, Ignacio Enrique; Optimal threshold estimation for binary classifiers using game theory; F1000 Research Ltd; F1000Research; 5; 2762; 25-11-2016; 1-12
2046-1402
CONICET Digital
CONICET
url http://hdl.handle.net/11336/60009
identifier_str_mv Sánchez Miguel, Ignacio Enrique; Optimal threshold estimation for binary classifiers using game theory; F1000 Research Ltd; F1000Research; 5; 2762; 25-11-2016; 1-12
2046-1402
CONICET Digital
CONICET
dc.language.none.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv info:eu-repo/semantics/altIdentifier/url/https://f1000research.com/articles/5-2762/v3
info:eu-repo/semantics/altIdentifier/doi/10.12688/f1000research.10114.3
dc.rights.none.fl_str_mv info:eu-repo/semantics/openAccess
https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
eu_rights_str_mv openAccess
rights_invalid_str_mv https://creativecommons.org/licenses/by-nc-sa/2.5/ar/
dc.format.none.fl_str_mv application/pdf
application/pdf
application/pdf
dc.publisher.none.fl_str_mv F1000 Research Ltd
publisher.none.fl_str_mv F1000 Research Ltd
dc.source.none.fl_str_mv reponame:CONICET Digital (CONICET)
instname:Consejo Nacional de Investigaciones Científicas y Técnicas
reponame_str CONICET Digital (CONICET)
collection CONICET Digital (CONICET)
instname_str Consejo Nacional de Investigaciones Científicas y Técnicas
repository.name.fl_str_mv CONICET Digital (CONICET) - Consejo Nacional de Investigaciones Científicas y Técnicas
repository.mail.fl_str_mv dasensio@conicet.gov.ar; lcarlino@conicet.gov.ar
_version_ 1844613954118090752
score 13.070432