HYBRID ARTIFICIAL NEURAL NETWORK AND FUZZY LOGIC FOR FUNCTION APPROXIMATION

Abstract

The problem of intelligent hybrid systems investigated in this study. Intelligent systems consist of fuzzy systems (FS) and neural networks (NN). This intelligent system has specific properties (modeling, ability of learning, obtaining empirical rules, solving optimizing tasks, classifying …) fitting certain type of applications. The combination of NN and FS systems makes fuzzy-NN system, neuron-fuzzy system. Such type of combination of systems is known as the hybrid intelligent systems (HIS). There are programs created in C++ and Matlab for these purposes, where many demo applications were made for different HIS in the area of system control and modeling. There are three programs have developed; Neural Network program (NNP), fuzzy program (FP) and Neural networks fuzzy program ( NNFP), to investigate the effect of these approaches on ANN learning using several datasets. The results have explored that Neural networks fuzzy (NNF) give quite better results in terms of small errors and convergence rate. compared to NN and FUZZY. The aim of the paper is to prove that the process of hybridization between the algorithms gives better results than the use of separate algorithms. This is known as the soft computing. This is implementation on the approximation functions.