High-Performance Convolutional Neural Network Emulation Via Fpga-Integrated Memristive Circuitry

Abstract

In this study, to tackle the challenges of high cost, insufficient reliability, and short lifespan associated with physical memristors, a FPGA-based memristor simulation circuit is proposed. This circuit is deployed on the Xilinx ZYNQ-7000 FPGA XC7Z020, with a resource utilization rate of less than 1%, and has been successfully integrated and validated. Moreover, the paper leverages a memristor model to build a convolutional neural network for MNIST image recognition, which achieves an accuracy of 90.22% with a parameter count of 512.

Keywords: memristor digital emulators digit recognition