A Method for KDD Based on Neural Network Ensembles

Abstract

Neural networks are important tools for KDD(Knowledge Discovery from Databases),but the knowledge derived with neural networks (especially the regression neural networks) is difficult to comprehend, and the training duration is rather long for mass data sets. In this paper, a method based on RBF neural networks ensembles is proposed for KDD. It is not only fast but also effective in using this method for extracting the rules from data.