The article is concerned with the data-based design of fuzzy systems with applications to medical diagnostic problems. Firstly, an overview about relevant approaches is given. Here, the focus is on tree-oriented methods applicable to the generation of fuzzy rules. In the second part, a new method is presented comprising the induction of decision trees and their translation into fuzzy rules. From these rules, new generalized rule hypotheses are generated and evaluated from which finally cooperating rules are selected into a rule base. The presentation of the method includes the application to a cardiologic diagnosis.
Print ISSN: 0178-2312
Volume: 48, 07/2000
Pages: 317