A Maximizing-Discriminability-Based Architecture for Fuzzy-Neural-Network Hardware

A maximizing-discriminability-based architecture for fuzzy-neural-network (FNN) hardware is proposed in this paper. The major contribution of this proposed FNN hardware is to increase the discriminative capability among different classes in classification problems by combining linear discriminant analysis (LDA) and Gaussian mixture model (GMM). In LDA, the weights are updated by seeking directions that are efficient for discrimination. In GMM, the parameter learning adopts the gradient descent method to reduce the cost function. Furthermore, this FNN can be reconfigured by the instruction of the external processer.