Abstract
A new methodology for the automatic generation and optimization of fuzzy rules is being proposed. The technique is based on the concepts of density based clustering, neural network and fuzzy logic. We develop the fuzzy rules from the given data using density based unsupervised clustering. The membership functions in the fuzzy rules are then developed. A fuzzy inference system is used to obtain the output. The fuzzy rules used in the inference mechanism, are optimized using back propagation to obtain the desired results. A well-known benchmark data is used to demonstrate the validity and applicability of the developed methodology.