COMPARISON OF THE MICHIGAN AND PITTSBURGH APPROACHES TO THE DESIGN OFFUZZY CLASSIFICATION SYSTEMS

Fuzzy systems based on fuzzy if-then rules have been applied to variou s problems. The main application area has been fuzzy control problems. In many cases, such fuzzy systems can handle only a few input variabl es. This is because the number of fuzzy if-then rules exponentially in creases as the number of input variables increases. In this gaper, we try to design fuzzy classification systems based on fuzzy if-then rule s for multidimensional pattern classification problems with many attri butes. For designing such fuzzy classification systems, we compare two frameworks in the area of genetics-based machine learning: the Michig an approach and the Pittsburgh approach. The performance of fuzzy rule -based classification systems is also compared with that of various pa ttern classification methods. In computer simulations, we use a wine c lassification problem with 13 attributes, a cancer diagnosis problem w ith 9 attributes, and a credit approval problem with 14 attributes. (C ) 1997 Scripta Technica, Inc.