Extended evaluation of the UPM method for multiclass problems

Multiclass problems are usually of high technological value, but many classification methods are binary in origin. In the last years, several improved solutions based on the combination of simple classifiers were introduced. An interesting solution is based on creating a hierarchy of sub-problems by...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Ahumada, Hernán César, Grinblat, Guillermo L., Granitto, Pablo Miguel
Formato: Objeto de conferencia
Lenguaje:Español
Publicado: 2011
Materias:
Ciencias Informáticas
UPM
Method for Multiclass Problems
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/125236
Aporte de:
SEDICI (UNLP) de Universidad Nacional de La Plata
id I19-R120-10915-125236
record_format dspace
institution Universidad Nacional de La Plata
institution_str I-19
repository_str R-120
collection SEDICI (UNLP)
language Español
topic Ciencias Informáticas
UPM
Method for Multiclass Problems
spellingShingle Ciencias Informáticas
UPM
Method for Multiclass Problems
Ahumada, Hernán César
Grinblat, Guillermo L.
Granitto, Pablo Miguel
Extended evaluation of the UPM method for multiclass problems
topic_facet Ciencias Informáticas
UPM
Method for Multiclass Problems
description Multiclass problems are usually of high technological value, but many classification methods are binary in origin. In the last years, several improved solutions based on the combination of simple classifiers were introduced. An interesting solution is based on creating a hierarchy of sub-problems by clustering prototypes of each one of the classes; there- fore the solution is heavily influenced by the label’s information. In this work we analyze a new strategy to solve multiclass problems that makes more use of spatial information than other methods. We construct a hier- archy of subproblems, but opposite to previous developments, based only on spatial information and not using a single prototype for each class. We evaluate the use of different clustering methods (either agglomera- tive or divisive) for this task and also the use two different classifiers (linear SVM and FDA–GenRidge) for each sub-problem (if needed, be- cause in several cases the clustering method directly gives a subset with samples of a single class). We compare the new method with several pre- vious approaches, finding promising results. The good performance of our approach is virtually independent of the classifier coupled to it, which suggest that it success is primarily related to the use of an appropriate clustering strategy.
format Objeto de conferencia
Objeto de conferencia
author Ahumada, Hernán César
Grinblat, Guillermo L.
Granitto, Pablo Miguel
author_facet Ahumada, Hernán César
Grinblat, Guillermo L.
Granitto, Pablo Miguel
author_sort Ahumada, Hernán César
title Extended evaluation of the UPM method for multiclass problems
title_short Extended evaluation of the UPM method for multiclass problems
title_full Extended evaluation of the UPM method for multiclass problems
title_fullStr Extended evaluation of the UPM method for multiclass problems
title_full_unstemmed Extended evaluation of the UPM method for multiclass problems
title_sort extended evaluation of the upm method for multiclass problems
publishDate 2011
url http://sedici.unlp.edu.ar/handle/10915/125236
work_keys_str_mv AT ahumadahernancesar extendedevaluationoftheupmmethodformulticlassproblems
AT grinblatguillermol extendedevaluationoftheupmmethodformulticlassproblems
AT granittopablomiguel extendedevaluationoftheupmmethodformulticlassproblems
bdutipo_str Repositorios
_version_ 1764820451424468993

Ejemplares similares