Knowledge Acquisition with Trainable Fuzzy Classification Systems

Abstract

This paper proposes a trainable fuzzy classification system based on fuzzy if-then rules. The proposed method subsequently modifies the grade of certainty of each fuzzy if-then rule by a simple error-correction leaning rule. That is, when a pattern is misclassified by a particular fuzzy if-then rule, the grade of certainty of that rule is reduced. In order to examine appropriate parameter specifications in the learning rule, we apply the proposed method to a two-class classification problem in a two-dimensional pattern space. Moreover we demonstrate the effectiveness of the proposed method by the application to commercial competitive analysis data.