Double-Weighting for Covariate Shift Adaptation

Supervised learning is often affected by a covariate shift in which the marginal distributions of instances (covariates $x$) of training and testing samples $p_\text{tr}(x)$ and $p_\text{te}(x)$ are different but the label conditionals coincide. Existing approaches address such covariate shift by ei...

ver descrição completa

Detalhes bibliográficos
Autores: Segovia, J.I, Mazuelas, S., Liu, A.
Formato: artículo
Estado:Versión publicada
Fecha de publicación:2023
País:España
Recursos:Basque Center for Applied Mathematics (BCAM)
Repositorio:BIRD. BCAM's Institutional Repository Data
OAI Identifier:oai:bird.bcamath.org:20.500.11824/1765
Acesso em linha:http://hdl.handle.net/20.500.11824/1765
Access Level:acceso abierto
Palavra-chave:Covariate Shift, Supervised Classification, Selection Bias, Minimax Classification
id ES_494c4e633bcd73f7c1fe95e82e3d92b0
oai_identifier_str oai:bird.bcamath.org:20.500.11824/1765
network_acronym_str ES
network_name_str España
repository_id_str
spelling Double-Weighting for Covariate Shift AdaptationSegovia, J.IMazuelas, S.Liu, A.Covariate Shift, Supervised Classification, Selection Bias, Minimax ClassificationSupervised learning is often affected by a covariate shift in which the marginal distributions of instances (covariates $x$) of training and testing samples $p_\text{tr}(x)$ and $p_\text{te}(x)$ are different but the label conditionals coincide. Existing approaches address such covariate shift by either using the ratio $p_\text{te}(x)/p_\text{tr}(x)$ to weight training samples (reweighted methods) or using the ratio $p_\text{tr}(x)/p_\text{te}(x)$ to weight testing samples (robust methods). However, the performance of such approaches can be poor under support mismatch or when the above ratios take large values. We propose a minimax risk classification (MRC) approach for covariate shift adaptation that avoids such limitations by weighting both training and testing samples. In addition, we develop effective techniques that obtain both sets of weights and generalize the conventional kernel mean matching method. We provide novel generalization bounds for our method that show a significant increase in the effective sample size compared with reweighted methods. The proposed method also achieves enhanced classification performance in both synthetic and empirical experiments.CNS2022-135203, “Early Prognosis of COVID-19 Infections via Machine Learning” funded by the AXA Research Fund202420242023info:eu-repo/semantics/articleinfo:eu-repo/semantics/publishedVersionapplication/pdfhttp://hdl.handle.net/20.500.11824/1765reponame:BIRD. BCAM's Institutional Repository Datainstname:Basque Center for Applied Mathematics (BCAM)Ingléshttps://proceedings.mlr.press/v202/segovia-martin23a/segovia-martin23a.pdfinfo:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023/CEX2021-001142-Sinfo:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica y Técnica y de Innovación 2017-2020/PID2019-105058GA-I00info:eu-repo/grantAgreement/Gobierno Vasco/ELKARTEK/info:eu-repo/grantAgreement/Gobierno Vasco/BERC/BERC.2022-2025Reconocimiento-NoComercial-CompartirIgual 3.0 Españahttp://creativecommons.org/licenses/by-nc-sa/3.0/es/info:eu-repo/semantics/openAccessoai:bird.bcamath.org:20.500.11824/17652026-06-19T12:47:47Z
dc.title.none.fl_str_mv Double-Weighting for Covariate Shift Adaptation
title Double-Weighting for Covariate Shift Adaptation
spellingShingle Double-Weighting for Covariate Shift Adaptation
Segovia, J.I
Covariate Shift, Supervised Classification, Selection Bias, Minimax Classification
title_short Double-Weighting for Covariate Shift Adaptation
title_full Double-Weighting for Covariate Shift Adaptation
title_fullStr Double-Weighting for Covariate Shift Adaptation
title_full_unstemmed Double-Weighting for Covariate Shift Adaptation
title_sort Double-Weighting for Covariate Shift Adaptation
dc.creator.none.fl_str_mv Segovia, J.I
Mazuelas, S.
Liu, A.
author Segovia, J.I
author_facet Segovia, J.I
Mazuelas, S.
Liu, A.
author_role author
author2 Mazuelas, S.
Liu, A.
author2_role author
author
dc.subject.none.fl_str_mv Covariate Shift, Supervised Classification, Selection Bias, Minimax Classification
topic Covariate Shift, Supervised Classification, Selection Bias, Minimax Classification
description Supervised learning is often affected by a covariate shift in which the marginal distributions of instances (covariates $x$) of training and testing samples $p_\text{tr}(x)$ and $p_\text{te}(x)$ are different but the label conditionals coincide. Existing approaches address such covariate shift by either using the ratio $p_\text{te}(x)/p_\text{tr}(x)$ to weight training samples (reweighted methods) or using the ratio $p_\text{tr}(x)/p_\text{te}(x)$ to weight testing samples (robust methods). However, the performance of such approaches can be poor under support mismatch or when the above ratios take large values. We propose a minimax risk classification (MRC) approach for covariate shift adaptation that avoids such limitations by weighting both training and testing samples. In addition, we develop effective techniques that obtain both sets of weights and generalize the conventional kernel mean matching method. We provide novel generalization bounds for our method that show a significant increase in the effective sample size compared with reweighted methods. The proposed method also achieves enhanced classification performance in both synthetic and empirical experiments.
publishDate 2023
dc.date.none.fl_str_mv 2023
2024
2024
dc.type.none.fl_str_mv info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
format article
status_str publishedVersion
dc.identifier.none.fl_str_mv http://hdl.handle.net/20.500.11824/1765
url http://hdl.handle.net/20.500.11824/1765
dc.language.none.fl_str_mv Inglés
language_invalid_str_mv Inglés
dc.relation.none.fl_str_mv https://proceedings.mlr.press/v202/segovia-martin23a/segovia-martin23a.pdf
info:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica y Técnica y de Innovación 2021-2023/CEX2021-001142-S
info:eu-repo/grantAgreement/AEI/Plan Estatal de Investigación Científica y Técnica y de Innovación 2017-2020/PID2019-105058GA-I00
info:eu-repo/grantAgreement/Gobierno Vasco/ELKARTEK/
info:eu-repo/grantAgreement/Gobierno Vasco/BERC/BERC.2022-2025
dc.rights.none.fl_str_mv Reconocimiento-NoComercial-CompartirIgual 3.0 España
http://creativecommons.org/licenses/by-nc-sa/3.0/es/
info:eu-repo/semantics/openAccess
rights_invalid_str_mv Reconocimiento-NoComercial-CompartirIgual 3.0 España
http://creativecommons.org/licenses/by-nc-sa/3.0/es/
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:BIRD. BCAM's Institutional Repository Data
instname:Basque Center for Applied Mathematics (BCAM)
instname_str Basque Center for Applied Mathematics (BCAM)
reponame_str BIRD. BCAM's Institutional Repository Data
collection BIRD. BCAM's Institutional Repository Data
repository.name.fl_str_mv
repository.mail.fl_str_mv
_version_ 1869407414119301120
score 15.300724

Registos relacionados