The application of fuzzy set and neural network in system identification and classification

Outlines a new learning algorithm to deal with interval information for a feedforward neural network. From the concept of fuzzy sets and using the vertex method, a new backpropagation neural network, called virtual vertex backpropagation, VVBP, is presented. A dynamic tracing machine is introduced to overcome weight correction in the learning process. Two applications in simple system identification and pattern classification are presented to discuss the effectiveness of VVBP.<
>