Study on the function of fuzzy data curve in fuzzy-neural network system
Zhifeng Jiang, Zemin Liu · 2002
This paper presents an improved fuzzy-neural network (FNN) model, which is a simple but effective fuzzy-rule based model about complex systems. Its architecture could be determined only by using the input-output data of the system. In the FNN model, the membership functions and weights would be adjusted optimally through the high speed BP learning principle, and the number of neurons in the inference layer would be confirmed through fuzzy data curves. Viewing the neural network as a fuzzy model gives an insight into the real system, and provides a method to simplify the neural network.