A Novel Radial Basis Function Neural Network For Approximation

Two difficulties are involved with traditional RBF networks: the initial configuration of an RBF network needs to be determined by a trial-and-error method, and the performance suffers degradation when the desired locations of the center of the RBF are not suitable. A novel RBF network is proposed to overcome these difficulties. A new radial basis function is used for hidden nodes, and the number of nodes is determined automatically by Shannon sampling theorem. The corresponding learning algorithm generally takes far less time for approximation with an optimized parameter setting. The locations of the centers of RBF are fixed. Experimental results have shown that the RBF networks constructed by our method have a smaller number of nodes, a faster learning speed, and a smaller approximation error than the networks produced by other methods.