COMPLETENESS AND CONSISTENCY CONDITIONS FOR LEARNING FUZZY RULES

The completeness and consistency conditions were introduced in order t o achieve acceptable concept recognition rules. In real problems, we c an handle noise-affected examples and it is not always possible to mai ntain both conditions. Moreover, when we use fuzzy information there i s a partial matching between examples and rules, therefore the consist ency condition becomes a matter of degree. In this paper, a learning a lgorithm based on soft consistency and completeness conditions is prop osed. This learning algorithm combines in a single process rule and fe ature selection and it is tested on different databases. (C) 1998 Else vier Science B.V. All rights reserved.