A method for learning of first-order cellular neural networks

In this paper, the first-order Cellular Neural Networks (CNN) with homogeneous weight structure are investigated and an approach to learn them is suggested. It is shown that all CNN weight templates are classified according to properties of possible stable states. As the result, the proposed learning method is based on the ideas of Perceptron learning rule. It allows to find parameters of a CNN connection template which provides the formation of patterns with preset properties. The method was applied to the 1D and the 2D CNN learning with symmetric templates and was verified by simulation.