Feedback GMDH-Type Neural Network Algorithm Using Prediction Error Criterion Defined as AIC

In this study, a feedback Group Method of Data Handling (GMDH)- type neural network algorithm using prediction error criterion defined as AIC, is proposed. In this algorithm, the optimum neural network architecture is automatically selected from three types of neural network architectures such as sigmoid function type neural network, radial basis function (RBF) type neural network and polynomial type neural network. Furthermore, the structural parameters such as the number of feedback loops, the number of neurons in the hidden layers and useful input variables are automatically selected so as to minimize the prediction error criterion defined as Akaike's Information Criterion (AIC). 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 identification problem of the complex nonlinear system. © Springer-Verlag Berlin Heidelberg 2012.