Fuzzy CMAC Neural Networks

Abstract

This article reveals the connection between the cerebellar model arithmetic computer (CMAC) neural network and fuzzy inference systems. A novel artificial neural network architecture called the fuzzy CMAC neural network is established that achieves the synergistic combination of the preferred features of the CMAC neural network and the fuzzy logic controller. The proposed combination of fuzzy logic and neural net techniques has significant potential to decrease the cost, complexity, and learning time over conventional neural nets. It retains the benefits of learning arbitrary functions through training, but it also captures the benefits of fuzzy logic, which allows heuristic rules to be generated by experts using linguistic variables of meaning to the particular problem domain. We believe that the work herein proposed will produce a generic fuzzy CMAC for a wide range of applications.