神经形态器件现状与未来

发布时间：2018-06-18 17:23

  本文选题：类脑计算 + 神经网络　； 参考：《国防科技》2016年06期

【摘要】：为解决现有冯·诺依曼计算机面临的低智能、高能耗、低容错等问题,"类脑计算"应运而生。"类脑计算"借鉴大脑的体系结构和信息处理方式,基于神经形态器件构建逻辑与存储相融合的计算硬件,旨在实现更为灵活和智能的信息处理与计算模式。在阐述"类脑计算"起源与发展现状的基础上,研究了人脑神经网络信息处理机制,重点介绍了神经形态器件的研发态势。神经形态器件作为类脑计算芯片和系统的基本组成单元,对于类脑智能的实现至关重要,因此研发能够高精度模拟生物突触、神经元信息处理功能的微纳器件是当前的研究热点。
[Abstract]:In order to solve the problems of low intelligence, high energy consumption and low fault tolerance faced by Von Neumann computer, "brain-like computing" came into being. In order to realize a more flexible and intelligent information processing and computing mode, brain-like computing "refers to the system structure and information processing mode of the brain, and constructs the computing hardware based on neural morphological devices to combine logic with storage." On the basis of expounding the origin and development of "brain like computing", the information processing mechanism of human brain neural network is studied, and the research and development trend of neural morphological devices is introduced emphatically. As the basic components of brain-like computing chips and systems, neuromorphic devices are very important for the realization of brain-like intelligence, so they can be developed to simulate biological synapses with high precision. The micro-nano devices with neuron information processing function are the focus of current research.
【作者单位】： 北京大学信息科学技术学院;
【分类号】：R318;TP183
，

本文编号：2036295