Fuzzy Expert System for Pattern Recognition

Abstract

In a pattern recognition environment it is often difficult to capture information in a precise form, and consequently, the knowledge about the recognition domain (problem) may be inexact, incomplete or not reliable. In order to properly represent the knowledge of this kind, fuzzy production rules have been used. In this paper fuzzy Petri nets are used as a knowledge representation scheme in which the fuzzy production rules are applied for describing a fuzzy relation between two propositions. Based on the fuzzy Petri net model of a knowledge base, an efficient algorithm is proposed to perform fuzzy reasoning automatically.