Simultaneous perturbation for single hidden layer networks -- cascade learning

A simultaneous perturbation approach for cascade learning of single hidden layer neural network is presented. A sigmoidal hidden neuron is added to the single layer of hidden neurons after training until the error has stopped decreasing after a certain limit. Then, the cascaded network is again trained using simultaneous perturbation. Perturbation employed on the weights connecting to hidden neurons are changed to detrap the local minima in training. The proposed technique gives better convergence results for the selected problems, namely neuro-controller, XOR, L–T character recognition, two spirals, simple interaction function, harmonic function and complicated interaction function. c © 2002 Elsevier Science B.V. All rights reserved.