Fuzzy If-Then Rules with Certainty Factors using Multilayer Model of GMDH

Abstract

In this paper, we proposed a method for automatically constructing fuzzy if-then rules with certainty factors from numerical data. In our method, fuzzy if-then rules are constructed only from numerical data. Our method requires no optimization techniques in generating of fuzzy if-then rules. It is easy to comprehend each of the constructed fuzzy if-then rules because they are represented by the certainty factors. Next we propose a method for applying the rule construction method to multi-layer models of GMDH where fuzzy if-then rules are used at each layer. Using the multi-layer models, it becomes easier to represent multi-input systems by fuzzy if-then rules.