IEICE Electronics Express
Online ISSN : 1349-2543
Quaternary Synapses Network for Memristor-based Spiking Convolutional Neural Networks
Sheng-Yang SunJiwei LiZhiwei LiHusheng LiuHaijun LiuQingjiang Li
Author information
JOURNALS FREE ACCESS Advance online publication

Article ID: 16.20190004

Details
Abstract

This paper proposes a method that renders the weights of the neural network with quaternary synapses map into the only four-level memristance of memristive devices. We show this method is capable of operating with a negligible loss in classification accuracy when the memristors utilized can store at least four unique values. Compared with other state-of-the-art methods, the method presented can achieve 98.65% accuracy under the 0.60M parameters. Systematic error analysis shows that the network can still reach over 95% accuracy under the condition of 95% yield of memristor crossbar array, 100 μV op-amp offset voltage and 0.5% Single-Pole-Double-Throw switches noise.

Information related to the author
© 2019 by The Institute of Electronics, Information and Communication Engineers
feedback
Top