Optimization of fuzzy inference rules by using the genetic algorithm and its application to the bond rating

Abstract

This paper deals with the optimization of membership function in the rules of fuzzy inference system by using the Genetic Algorithm, and also shows its application to the classification of corporate bond. In the fuzzy inference system the parameters such as the weight are determined to minimize the difference between the prescribed value and the output of the system. However, the output of the system also depends on the shape of the membership function. We utilize the Genetic Algorithm to select better shape of the membership function. The method is applied to the automatic classification of corporate bond (so called bond rating) of Japanese firms. The result shows about 5% improvement of the bond rating compared to the conventional fuzzy inference system.