SECA: a stepwise algorithm for construction of neural networks ensembles

Ensembles of artificial neural networks (ANN) have been used in the last years as classification/regression machines, showing improved generalization capabilities that outperform those of single networks. However, it has been recognized that for aggregation to be effective the individual networks mu...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Granitto, Pablo Miguel, Verdes, Pablo Fabián, Ceccatto, Hermenegildo Alejandro, Navone, Hugo Daniel
Formato: Objeto de conferencia
Lenguaje:Inglés
Publicado: 2001
Materias:
Ciencias Informáticas
Neural nets
Algorithms
ARTIFICIAL INTELLIGENCE
machine learning
ensemble methods
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/23398
Aporte de:
SEDICI (UNLP) de Universidad Nacional de La Plata
id I19-R120-10915-23398
record_format dspace
institution Universidad Nacional de La Plata
institution_str I-19
repository_str R-120
collection SEDICI (UNLP)
language Inglés
topic Ciencias Informáticas
Neural nets
Algorithms
ARTIFICIAL INTELLIGENCE
machine learning
ensemble methods
spellingShingle Ciencias Informáticas
Neural nets
Algorithms
ARTIFICIAL INTELLIGENCE
machine learning
ensemble methods
Granitto, Pablo Miguel
Verdes, Pablo Fabián
Ceccatto, Hermenegildo Alejandro
Navone, Hugo Daniel
SECA: a stepwise algorithm for construction of neural networks ensembles
topic_facet Ciencias Informáticas
Neural nets
Algorithms
ARTIFICIAL INTELLIGENCE
machine learning
ensemble methods
description Ensembles of artificial neural networks (ANN) have been used in the last years as classification/regression machines, showing improved generalization capabilities that outperform those of single networks. However, it has been recognized that for aggregation to be effective the individual networks must be as accurate and diverse as possible. An important problem is, then, how to tune the aggregate members in order to have an optimal compromise between these two conflicting conditions. Recently, we proposed a new method for constructing ANN ensembles —termed here Stepwise Ensemble Construction Algorithm (SECA)— which leads to overtrained aggregate members with an adequate balance between accuracy and diversity. We present here a more extensive evaluation of SECA and discuss a potential problem with this algorithm: the unfrequent but damaging selection through its heuristic of particularly bad ensemble members. We introduce a modified version of SECA that can cope with this problem by allowing individual weighing of aggregate members. The original algorithm and its weighed modification are favorably tested against other methods, producing an improvement in performance on the standard statistical databases used as benchmarks.
format Objeto de conferencia
Objeto de conferencia
author Granitto, Pablo Miguel
Verdes, Pablo Fabián
Ceccatto, Hermenegildo Alejandro
Navone, Hugo Daniel
author_facet Granitto, Pablo Miguel
Verdes, Pablo Fabián
Ceccatto, Hermenegildo Alejandro
Navone, Hugo Daniel
author_sort Granitto, Pablo Miguel
title SECA: a stepwise algorithm for construction of neural networks ensembles
title_short SECA: a stepwise algorithm for construction of neural networks ensembles
title_full SECA: a stepwise algorithm for construction of neural networks ensembles
title_fullStr SECA: a stepwise algorithm for construction of neural networks ensembles
title_full_unstemmed SECA: a stepwise algorithm for construction of neural networks ensembles
title_sort seca: a stepwise algorithm for construction of neural networks ensembles
publishDate 2001
url http://sedici.unlp.edu.ar/handle/10915/23398
work_keys_str_mv AT granittopablomiguel secaastepwisealgorithmforconstructionofneuralnetworksensembles
AT verdespablofabian secaastepwisealgorithmforconstructionofneuralnetworksensembles
AT ceccattohermenegildoalejandro secaastepwisealgorithmforconstructionofneuralnetworksensembles
AT navonehugodaniel secaastepwisealgorithmforconstructionofneuralnetworksensembles
bdutipo_str Repositorios
_version_ 1764820465881186305

Ejemplares similares