THE APPROXIMATION ABILITY OF FUZZY NEURAL NETWORKS
Zhao Mingjie · 2000
A new theorem on an integral transform of multi variable functions is presented based on the multi variable Fourier transform theory. Using this theorem, the relationship between the approximation error and the structure of the fuzzy function approximator is discussed. And the approximation accuracy, which is inversely proportional to the number of hidden units of the fuzzy function approximator, is obtained. This result shows that the approximation accuracy is independent of the input dimensions.