A discrete memristive neural network and its application for character recognition

Design of artificial neural networks based on memristor has attracted increasing attentions from researchers. However, there are no reports on the discrete memristor based neural network. In this work, a novel discrete memristor BP neural network is designed. Firstly, a discrete memristor is introduced, in which the internal state can be controlled by the input discrete current signal. Secondly, theoretical basis of the proposed discrete memristor BP neural network is given where the weights are defined by the memristor resistance and can be adjusted according to discrete feedback errors. Finally, a three layers discrete memristor BP neural network is built to perform MNIST-10 handwriting recognition. Numerical results show a high classification accuracy of 97.16% and it verifies the effectiveness of the proposed method.

Automated summary

The design of artificial neural networks based on memristors has received increasing attention from researchers. However, there have been no reports of discrete memristor-based neural networks. In this study, a novel discrete memristor BP neural network is designed. A three-layer discrete memristor BP neural network is built for MNIST-10 handwriting recognition, achieving a high classification accuracy of 97.16%, which demonstrates the effectiveness of the proposed method.