相变存储器及其用于神经形态计算的研究综述

目前随着人工智能领域的兴起以及人们对数据存储和计算的强烈需求,迫切需要存储器的改进和类似于人脑的高效存储运算效率。所以,相变存储器及其用于神经形态计算的研究是极具价值的。相变存储材料(PCMs)受到激发时所产生的电阻值变化可以用来建立尖峰神经网络从而实现模拟神经形态计算系统。本文介绍了相变存储器物理机制,其中包括相变材料的相变原理及主要性能特征,重点叙述了相变存储器在优化存储与计算方向的研究进展和应用,进而为该领域未来的发展方向提供参考。

Review of Phase Change Memory and Its Application in Neuromorphic Computation

With the rise of artificial intelligence and the explosive demand for data storage and computing, the improvement of memory and the efficient storage and computing efficiency similar to that of human brain are urgently needed. Therefore, it is necessary to study phase change memory and its application in neuromorphic computation. The change of resistance caused by the excitation of phase change materials (PCMs) can be used to build the spike neural network and realize the simulation of neural morphological computing system. This paper introduces the physical mechanism of phase change memory, including the phase change principle and main performance characteristics of phase change materials, and focuses on the research progress and application of phase change memory, so as to provide reference for the future development direction of this field.