Chaos associative memory model

In this paper we propose a chaos dynamic memory model applied to a chaotic autoassociation memory. The present artificial neuron model is properly characterized in terms of a time-dependent sinusoidal activation function to involve transient chaotic dynamics as well as the energy steepest descent strategy. It is shown that the present neural network has a remarkable retrieval ability beyond that of conventional models with such a monotonous activation function as a sigmoidal one. This advantage is found to result from the analogue periodic mapping accompanied by the chaotic behaviour of the neurons as well as the symmetry of the dynamic equation.