A fast learning algorithm for parsimonious fuzzy neural systems
Shiqian Wu, Meng Joo Er · 1999
In this paper, a fast learning algorithm for Dynamic Fuzzy Neural Networks (D-FNNs) based on extended Radial Basis Function (RBF) neural networks, which are functionally equivalent to TSK fuzzy systems, is proposed. The algorithm has fast learning speed and dynamic self-organizing structure. A parsimonious system can be achieved based on a new pruning technology called Error Reduction Ratio (ERR). Simulation studies and comparisons with some other learning algorithms demonstrate that the proposed algorithm is superior.