NEURAL NETWORK IMPLEMENTATION OF FUZZY-LOGIC

Fuzzy logic has gained increased attention as a methodology for managing uncertainty in a rule-based structure. In a fuzzy logic inference system, more rules can fire at any given time than in a crisp expert system. Since the propositions are modelled as possibility distributions, there is a considerable computation load on the inference engine. In this paper, a neural network structure is proposed as a means of performing fuzzy logic inference. Three variations of the network are described, but in each case, the knowledge of the rule (i.e., the antecedent and consequent clauses) are explicitly encoded in the weights of the net. The theoretical properties of this structure are developed. In fact, the network reduces to crisp modus ponens when the inputs are crisp sets. Also, under suitable conditions the degree of specificity of the consequence of the inference is a monotone function of the degree of specificity of the input. Several simulation studies are included to illustrate the performance of the fuzzy logic inference networks.