Feedback GMDH-type neural network algorithm using prediction error criterion for self-organization

Abstract

In this study, a feedback group method of data handling (GMDH)-type neural network algorithm using prediction error criterion for self-organization is proposed. In this algorithm, the optimum neural network architecture is automatically selected from three types of neural network architectures such as the sigmoid function type neural network, the radial basis function (RBF) type neural network and the polynomial type neural network. Furthermore, the structural parameters such as the number of feedback loops, the number of neurons in the hidden layers and the useful input variables are automatically selected so as to minimize the prediction error criterion defined as Akaikepsilas information criterion (AIC) or prediction sum of squares (PSS). The feedback GMDH-type neural network has a feedback loop and the complexity of the neural network increases gradually using feedback loop calculations so as to fit the complexity of the nonlinear system. This algorithm is applied to the identification problem of the complex nonlinear system.