Rule extraction through fuzzy modeling using fuzzy neural network

Presents a rule extraction method from data using fuzzy neural networks (FNNs) and a genetic algorithm (GA). This method is based on a new framework for fuzzy modeling. This framework consists of a hypothesis generation block and a hypothesis evaluation block. The generation block, using GA, searches for the set of rules by generating candidates and the evaluation block guides the direction of the GA search. The FNN is used to fine tune the obtained fuzzy rules. A numerical experiment is done to show the feasibility of the proposed method.