Parameters optimization of T-S fuzzy classification system using PSO and SVM

In this paper, Takagi-Sugeno fuzzy classification system (T-S FCS) using particle swarm optimization (PSO) and support vector machine (SVM) for parameters optimization is proposed. The T-S FCS is constructed by fuzzy if-then rules whose consequents are linear state equations. The antecedents of T-S FCS are determined by the fuzzy membership of the input feature vectors. The prespecified values during the antecedent construction process are further optimized by using PSO. Consequent parameters in T-S FCS are learned through SVM. The proposed T-S FCS is able to minimize the effect of uncertainties, reduce the influence of artificial factors and give the system better generalization performance, which inherits the benefits of T-S fuzzy system, PSO and SVM. For demonstration, T-S FCS is used as a classifier in gender recognition. Comparisons with other mainstream classifiers, the advantages of the proposed T-S FCS are verified by experimental results.