Dynamical Properties of Dense Associative Memory

Abstract: The dense associative memory is one of the basic modern Hopfield networks and can store large numbers of memory patterns. While the stationary state storage capacity has been investigated so far, its dynamical properties have not been discussed. In this paper, we analyze the dynamics by means of an exact approach based on generating functional analysis. It allows us to investigate convergence properties as well as the size of the attraction basins. We also analyze the stationary state of the updating rule.