The cross-coupled network converging the state to the global minimum with temporal increase of connection weights

Abstract

The authors have already proposed that the internal value of the cross coupled network can work as a similarity measure between a state of the network and one of patterns embedded in the network. The measure was applied to design a dynamic of a novel activation. Moreover, we proposed a nonequilibrium cross-coupled network which works as a dynamical associative memory. In this paper, we apply the similarity measure, that is the internal value, to achieve convergence of the state of the network to the global minimum.