General models of artificial neural networks

Abstract

The authors present a unified approach to the development of general forms of artificial neural network (ANN) models containing all well-known ANN models, or a majority of them. Starting with nonlinear dynamic models of n-neurons, and using the concept of signal-distribution matrices, the general forms of ANN models, as continuous and discrete-time nonlinear systems, are derived. All well known ANN models, like the Hopfield model, the McCullough and Pitts model, the linear LSS model, a multilayered feedforward model, and so on, can be obtained by using the general forms of ANN models. >