Analog neural network based on memristor crossbar arrays

Abstract

In this paper, a new feed forward analog neural network is designed using a memristor based crossbar array architecture. This structure consists of positive and negative polarity connection matrices. In order to show the performance and usefulness of the proposed circuit, it is considered a sample application of iris data recognition. The proposed neural network implementation is approved by the simulation in Cadence design environment using 0.35µm CMOS technology. The results obtained are promising for the implementation of high density neural network.