Time–Adaptive Support Vector Machines

In this work we propose an adaptive classification method able both to learn and to follow the temporal evolution of a drifting concept. With that purpose we introduce a modified SVM classifier, created using multiple hyperplanes valid only at small temporal intervals (windows). In contrast to oth...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Grinblat, Guillermo, Granitto, Pablo M., Ceccatto, Alejandro
Formato: Artículo
Lenguaje:en_US
Publicado: Asociación Española de Inteligencia Artificial 2011
Materias:
Adaptive methods
Support Vector Machine
Drifting concepts
Acceso en línea:http://hdl.handle.net/2133/1718
http://hdl.handle.net/2133/1718
Aporte de:
Repositorio Hipermedial de la Universidad Nacional de Rosario (UNR) de Universidad Nacional de Rosario
id I15-R121-2133-1718
record_format dspace
institution Universidad Nacional de Rosario
institution_str I-15
repository_str R-121
collection Repositorio Hipermedial de la Universidad Nacional de Rosario (UNR)
language en_US
orig_language_str_mv en_US
topic Adaptive methods
Support Vector Machine
Drifting concepts
spellingShingle Adaptive methods
Support Vector Machine
Drifting concepts
Grinblat, Guillermo
Granitto, Pablo M.
Ceccatto, Alejandro
Time–Adaptive Support Vector Machines
topic_facet Adaptive methods
Support Vector Machine
Drifting concepts
description In this work we propose an adaptive classification method able both to learn and to follow the temporal evolution of a drifting concept. With that purpose we introduce a modified SVM classifier, created using multiple hyperplanes valid only at small temporal intervals (windows). In contrast to other strategies proposed in the literature, our method learns all hyperplanes in a global way, minimizing a cost function that evaluates the error committed by this family of local classifiers plus a measure associated to the VC dimension of the family. We also show how the idea of slowly changing classifiers can be applied to non-linear stationary concepts with results similar to those obtained with normal SVMs using gaussian kernels.
format Article
author Grinblat, Guillermo
Granitto, Pablo M.
Ceccatto, Alejandro
author_facet Grinblat, Guillermo
Granitto, Pablo M.
Ceccatto, Alejandro
author_sort Grinblat, Guillermo
title Time–Adaptive Support Vector Machines
title_short Time–Adaptive Support Vector Machines
title_full Time–Adaptive Support Vector Machines
title_fullStr Time–Adaptive Support Vector Machines
title_full_unstemmed Time–Adaptive Support Vector Machines
title_sort time–adaptive support vector machines
publisher Asociación Española de Inteligencia Artificial
publishDate 2011
url http://hdl.handle.net/2133/1718
http://hdl.handle.net/2133/1718
work_keys_str_mv AT grinblatguillermo timeadaptivesupportvectormachines
AT granittopablom timeadaptivesupportvectormachines
AT ceccattoalejandro timeadaptivesupportvectormachines
bdutipo_str Repositorios
_version_ 1764820410587676675

Ejemplares similares