Neural Network Training Using a GMDH Type Algorithm

Authors have developed a Group Method of Data Handling (GMDH) type algorithm for designing multilayered neural networks. The algorithm is general enough that it will accept any number of inputs and any sized training set. Each neuron of the resulting network is a function of two of the inputs to the layer. The equation for each of the neurons is a quadratic polynomial. Several forms of the equation are automatically tested for each neuron to make sure that only the best equation of two inputs is kept. All possible combinations of two inputs to each layer are also tested by the algorithm and only the best neurons at each level are retained. The algorithm’s goal is to create as accurate network as possible while minimizing the size of the network. Software was developed to train and simulate networks using our algorithm. Several applications were modeled using our software, and the results demonstrate that our algorithm succeeded in developing small, accurate and multi-layer networks. This paper further presents the hardware implementation of the algorithm that consumes less power than other such systems.