A Novel Fuzzy Classifier Ensemble System

In this paper, a novel fuzzy classifier ensemble system is proposed. This system can reduce subjective factor in building a fuzzy classifier, and improve the classification recognition rate and stability. Three proposed approaches are introduced, namely, the approach of measuring generalization difference(GD) of classifier sets to select individual classifiers, the approach of determining individual classifier´s reliability by the proposed membership matrix, the approach of classifier ensemble. The proposed system is evaluated with standard data sets. The comparison of experiments and the existed classifier ensemble systems. The experiment results show that the recognition rate of our proposed system is higher than ones of other classifier ensemble systems.