Approaches of Individual Classifier Generation and Classifier Set Selection for Fuzzy Classifier Ensemble

Classifier ensemble is now an active area of research in machine learning and pattern recognition. Fuzzy classification is an important application of fuzzy set. In this paper, we propose a fuzzy classifier with kenerl fuzzy C-means clustering (KFCMC) algorithm. Based on such fuzzy classifier, the approaches of constructing fuzzy classifier ensemble system are introduced. These approaches include individual fuzzy classifier generation, individual fuzzy classifier reliability computation, fuzzy classifier set selection, and classifiers ensemble etc. Our aim is building accurate and diverse classifiers. Experiment results show that our proposed approaches are effective.