I. Introduction
Recent advances in the development of artificial neural networks (ANNs) including the convolutional neural network, recursive neural network and other deep neural networks significantly extend the capability of artificial intelligence in various tasks such as image recognition, natural language processing, self-driving cars, etc. [1]–[3]. However, the operation of these ANNs with the modern computer systems based on the conventional von Neumann architecture is seriously restricted due to the fact that the memory and central processing units are physically separated. Brain-inspired neuromorphic computing based on a non-von Neumann architecture becomes an emerging field for the efficient implementation of artificial intelligence [4]–[6].