Dynamic Learning and Retrieving Scheme Based on Chaotic Neuron Model

Abstract

A stimulus-response scheme is introduced in chaotic neural networks with synaptic plasticities and the processes of dynamic learning under external stimuli are investigated. Owing to the refractoriness and the time-hysteresis, memory fixing abilities of stimuli become much higher than those by the Hopfield neural network with/without stochastic activities and also have sensitive dependences on the strength of stimulation. These characteristics turn out to be supported with the chaotic activity by examining the relation between the refractoriness and the Lyapunov exponent during the engraving of stimuli on the network. The above results indicate a possibility of realizing the real-time learning mechanism against the external time series inputs, which is difficult for the static Hopfield model of associative memory. Furthermore, modeling of dynamic retrieval mechanism is attempted for diverse switching phenomena on the chaotic neural network with grown (fixed) synapses.