IDENTIFICATION OF FUZZY RELATIONAL EQUATIONS BY FUZZY NEURAL NETWORKS

Although the identification of a fuzzy system by a feedforward neural network is obviously possible, whenever the system was continuous, however when we use fuzzy max-min neural network, some problems on the learning of the network arise. In this paper we present a method to identify a fuzzy relation by a fuzzy max-min neural network. We adapt the learning algorithm backpropagation to learning of max-min neural network by using one which we will denote as ''smooth derivative''. Finally, we present some examples and a comparison with a similar method.