Meysam Orouskhani - Ph. D. student, Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran,
Mohammad Mansouri - Ph. D. student, Intelligent System Laboratory, Electrical and Computer Engineering Department, K.n.Toosi University, Tehran, Iran,
Mohammad Teshnehlab - Academic member, Intelligent System Laboratory, Electrical and Computer Engineering Department, K.n.Toosi University, Tehran, Iran,

This paper introduces a new approach for training the Tsukamoto-Type neural fuzzy inference network (TNFIN). In the standard method, the antecedent and consequent parameters are trained by a hybrid learning algorithm combining the Least Square Estimation (LSE) method and the Gradient Descent (GD) method. In this study in order to tune the parameters of TNFIN, a new swarm-based optimization algorithm is applied. Cat Swarm Optimization as a novel swarm intelligence algorithm which is used for global optimization problems is used to train nonlinear parameters of TNFIN. Experimental result for prediction of Mackey-Glass model and identification of a nonlinear dynamic system indicates that the performance of proposed algorithm in comparison with standard method is much better and it shows quite satisfactory results.

Cat Swarm Optimization, Tsukamoto Fuzzy Model, Prediction and Identification