A Comparative Analysis of Fuzzy Model and Neural Network Model by Plural Performance Indices(Journal of Japan Society for Fuzzy Theory and Systems)

概要

論文の詳細を見る
Both fuzzy model and neural network model can be regarded as the method to represent nonlinear input/output relation of multi input and multi output system. But there are few researches concerning on what model is best in view of incorporating those models into microcomputer based systems. In this paper, we introduce plural performance indices of models such as model error, memory capacity necessary for implementation, learning speed, execution speed, etc., which designer must consider for implementing intelligent systems. We analyze and compare basic properties of fuzzy model and neural network model based on these plural performance indices. First of all, we introduce the notion of perfectly optimal model and noninferior models as rational candidate models with plural performance indices, which are obtained by applying the perfectly optimal solution and noninferior solution treated in the multi-objective decision problem. Furthermore, when there does not exist the perfectly optimal model, we discuss the two types of methods to select a preference optimal model from among the noninferior models obtained by checking the noninferiority of the models, according to the designer's preference or the requirements imposed on the designing process. The first method is based on the information criteria, and the second one is a kind of interactive satisfaction methods where requirements on the plural performance values are gradually obtained through the interactive process. Numerical examples are given to illustrate the effectiveness of the above two kinds of the methods.
日本知能情報ファジィ学会の論文
1992-10-15

A Comparative Analysis of Fuzzy Model and Neural Network Model by Plural Performance Indices(Journal of Japan Society for Fuzzy Theory and Systems)

スポンサーリンク

概要

著者

関連論文

スポンサーリンク