Cellular Neural Computations. The Formal Model and Possible Applications

Abstract

A formal model of mass parallel computations (the cellular neural algorithm) that contains features of the parallel substitution algorithm (a generalization of the von Neumann cellular automaton) and artificial neural networks is proposed. A formal representation of the cellular neural algorithm is given. Three typical examples are considered: the recovery of a distorted pattern, the search for the shortest path on a graph, and the selection of a contour in an image.