Increase the number of stable equilibrium points in a Hopfield-type neural network

We consider a Hopfield-type continuous neural network with feedbacks when this network has many stable and unstable equilibrium points and is used as an associative memory device. The purpose of the paper is to suggest a new method to increase the number of stable equilibrium points in this network, as the number of such points is one of the most important working characteristics of the network and represents the storage capacity for associative memories. On the basis of the initial artificial neural network system we generate a new system in which all the original equilibrium points (stable and unstable) have become stable absolute minima. Associative memory devices with more storage capacity can be designed as a result.