Support-Vector-Based Fuzzy Neural Networks

Abstract

In this paper, novel fuzzy neural networks (FNNs) combining with support vector learning mechanism called support-vector-based fuzzy neural networks (SVFNNs) are proposed for pattern classification and function approximation. The SVFNNs combine the capability of minimizing the empirical risk (training error) and expected risk (testing error) of support vector learning in high dimensional data spaces and the efficient human-like reasoning of FNN in handling uncertainty information. A learning algorithm consisting of three learning phases is developed to construct the SVFNNs and train the parameters. In the first phase, the fuzzy rules and membership functions are automatically determined by the clustering principle. In the second phase, the parameters of FNN are calculated by the SVM and SVR with the proposed adaptive fuzzy kernel function for pattern classification and function approximation, respectively. In the third phase, the relevant fuzzy rules are selected by the proposed fuzzy rule reduction method. To investigate the effectiveness of the proposed SVFNNs, they are applied to the Iris and Vehicle datasets for classification, and oneand twovariable functions for approximation, respectively. Experimental results show that the proposed SVFNNs can achieve good pattern classification and function approximation performance with drastically reduced number of fuzzy kernel functions (fuzzy rules).