BDBComp
Parceria:
SBC
On the Use of Fuzzy ART and SOM Networks in Ensemble Classifiers: A Performance Comparison

César L. C. MattosGuilherme A. Barreto

In this paper we introduce two novel ensemble models that are built using Fuzzy ART (FA) and SOM networks as base classifiers. For this purpose, we first describe three different strategies to convert these unsupervised com- petitive learning algorithms to supervised ones to allow them to be applied to pattern classification tasks. Then, a metaheuristic solution based on a hybrid PSO algorithm is devised for parameter optimization of the proposed ensemble classifiers. A comprehensive performance comparison using 10 benchmarking data sets indicates that the FA- and SOM-based ensemble classifiers consistently outperform ensembles built from standard supervised neural networks, such as the Fuzzy ARTMAP and the Extreme Learning Machine.

http://www.lbd.dcc.ufmg.br/colecoes/enia/2011/003.pdf

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato: bdbcomp@lbd.dcc.ufmg.br
     Mantida por:
LBD