GMDH-type neural networks: with radial basis functions and their application to medical image recognition of the brain

Abstract

In this paper, the group method of data handling (GMDH)-type neural networks with radial basis functions are proposed. Such networks can automatically organize themselves by using a heuristic self-organization method. In this algorithm, the network architecture can be automatically adjusted according to the complexity of the approximated nonlinear system. The number of hidden layers and the number of neurons in the hidden layers are selected so as to minimize an error criterion defined as Akaike's information criterion (AIC). Furthermore, various types of nonlinear combinations of variables are initially generated in each layer and only the useful combinations are selected by using AIC. In this study, the GMDH-type neural networks with radial basis functions are applied to medical image recognition of the brain. It is shown that this algorithm is simple and useful in medical image recognition of the brain.