Ahatova

In this article the authors analyze the use of expert systems for solving diagnostic problems in many subject areas of human activity. The need for use of automated approach to formatting a knowledge base of the expert system based on the fuzzy neural network is substantiated. The need for developing a special hybrid algorithm based on the back propagation algorithm and the genetic algorithm in order to increase the efficiency of its learning is actualized. The authors provide a description of the developed algorithm. It was experimentally shown that application of this algorithm can improve the speed and accuracy of adjustment of parameters of membership functions of antecedents of fuzzy rules derived from teaching a fuzzy neural network. By the example of the task of medical diagnostic, efficiency of the proposed approach used for formatting knowledge bases of expert systems is shown.