Neural Network Ensemble Based on Feature Selection

Abstract

In this paper, a novel neural network ensemble model, i.e. NNEIPCABag, which combines the feature selection technique, the improved principal component analysis (IPCA), with the Bagging method, is presented. Then the proposed model is employed for time series forecasting with the favor results obtained, which shows that the generalization ability of the proposed model can be superior to that of neural network ensemble only with the Bagging method, .i.e. NNEBag.