DAVION LEAST SQUARES-BASED LEARNING ALGORITHM FOR FEEDFORWARD NEURAL NETWORKS

This paper presents a new learning methodology for feedforward neural networks. The proposed algorithm is based on Davidon's least squares m inimization approach. The performance of the Davidon algorithm is comp ared with that of the back propagation for three prototype examples. T hese examples are: Case I, a linear second-order system; Case II, a ch aotic system; and Case III, a nonlinear dynamic system. The trained ne twork is employed to determine the one-step-ahead prediction of the ou tput of a given system. The simulation results show that, in most case s, the Davidon algorithm has an order of magnitude faster convergence rate than that of the back propagation. In order to make a fair compar ison, an optimum back propagation learning rate is used. The learning rate chosen is the one that results in the fastest convergence for a g iven system.