Evolutionary Design of Neuro-Fuzzy Systems - A Generic Framework

Abstract

A Neuro-Fuzzy (NF) system is a combination of Artificial Neural Network (ANN) and Fuzzy Inference System (FIS) in such a way that ANN learning algorithms are used to determine the parameters of FIS. Potential interactions between connectionist learning systems and evolutionary search procedures have attracted lots of research work recently. There is no guarantee that the learning algorithm converges in ANN and the tuning of FIS will be successful. Success of evolutionary search procedures for optimal design of ANNs and FIS are well proven and established in many application areas. However, the evolutionary design of a NF system is yet to be explored. In this paper, we attempt to formulate a 5-tier hierarchical evolutionary search procedure for the optimal design of NF systems. We formulate each of the evolutionary search procedures in detail and the interactions among them.