Fuzzy Logic Based Neural Network for Case Based Reasoning

Abstract

This chapter provides a way of integrating the concepts of neuro-fuzzy computing and case based reasoning for designing an efficient decision support system. Multilayer networks with fuzzy AND-neurons as hidden nodes and fuzzy OR-neurons as output nodes are used for this purpose. Lingustic fuzzy sets are considered at the input level. Cases are described as fuzzy IF-THEN rules in order to handle imprecision and vagueness. Relations of the weighting factors antecedents of rules, the certainty factors and the network parameters are analyzed. The effectiveness of the system is demonstrated on various synthetic and real life problems including the area of telecommunication.