Identification of the Multi-layered Neural Networks by Revised GMDH-Type Neural Network Algorithm with PSS Criterion

Abstract

In this paper, the revised GMDH-type neural network algorithm with PSS criterion for model selection is proposed. In this algorithm, the optimum multi-layered neural network architecture is automatically organized so as to minimize the prediction error criterion defined as PSS (Prediction Sum of Squares) by using the heuristic self-organization method. Both the sigmoid function type neural networks and the radial basis function type neural networks can be organized by this algorithm and the structural parameters such as the number of neurons in each layer, the number of layers and the useful input variables are automatically determined by using PSS criterion. Therefore, it is easy to apply this algorithm to the identification problem of the complex nonlinear system and to obtain a good prediction accuracy.