A New Model of Associative Memory Neural Network Based on An Improved Memristor

Abstract

In order to simulate biological associative learning, extensive studies have been conducted on associative memory networks. This paper designs a learning forgetting model with an improved forgetting memristor based on associative memory neural network and the corresponding circuit implementation is presented. The entire circuit includes a neuron module and a synapse module. The neuron module circuit has two inputs and one output. The synapse module circuit is mainly composed of resistance and memristor. In complete circuit, associative learning and the three forgetting processes together form a full functional associative memory model. Then, the new designed memory model reflects that people learn faster during secondary learning by adjusting the speed of memory. Besides, this model is able to take longer to forget during secondary forgetting, which can simulate young people’s and elderly people’s abilities to learn and forget. The simulation results of PSPICE show the effectiveness of the designed circuit. The memory model proposed in this article provides the possibility to better understand the memory function of human brain.