Multi-objective optimization of TSK fuzzy models

In this paper we propose a hybrid algorithm to optimize the structure of TSK type fuzzy model using back-propagation (BP) learning algorithm and non-dominated sorting genetic algorithm (NSGA-II). In a first step, BP algorithm is used to optimize the parameters of the model (parameters of membership functions and fuzzy rules). NSGA-II is used in a second phase, to optimize the number of fuzzy rules and to fine tune the parameters. A well known benchmark is used to evaluate performances of the proposed modeling approach, and compare it with other modeling approaches.