Sixth International Conference on Hybrid Intelligent Systems (HIS'06)
Ensemble as a Piecewise Linear Classifier
Auckland, New Zealand
December 13-December 15
ISBN: 0-7695-2662-4
In this paper we analyze the performance of a neural network ensemble in performing piecewise linear classification by automatically decomposing a non-linear problem into several linear sub-problems. The strength and weakness of this neural network ensemble with respect to MLP and Perceptron, and the diversity in the ensemble?s modules, created as the result of the competitive learning process, are the main focus of this paper.