Modeling the restoration of biological and biotechnical systems using hardware analog and software artificial neural networks

Abstract

A hardware analog model of an artificial neural network was developed, based on a specially trained software artificial neural network, for modeling the process of recovering damaged biological and biotechnical systems using neurochips based on the evolutionary method of training. A series of 12 computational experiments on the restoration of a damaged hardware analog artificial neural network with the help of a software artificial neural network was carried out. To restore a damaged network, an evolutionary approach is used. In most cases, it is possible to restore a damaged hardware analog neural network to 100% accuracy. The obtained results confirm the efficiency of the proposed approach in the framework of modeling the restoration of damaged biological and biotechnical systems using a neurochipon the basis of the evolutionary method using the "isolation" mechanism. The proposed recovery method opens up prospects for such areas as neuroprosthetics, self-learning and self-adapting systems; reverse-engineering; restoration of damaged data banks, image restoration; decision making and management, and so on.