忆阻器基神经形态计算器件与系统研究进展  

Research Progress of Memristor-Based Neuromorphic Computing Devices and Systems

作　　者：郭文斌 汪泽清 吴祖恒 蔺智挺[1,2] GUO Wenbin;WANG Zeqing;WU Zuheng;LIN Zhiting(School of Integrated Circuits,Anhui University,Hefei 230601,China;High Performance Integrated Circuit Engineering Research Center of Anhui Province,Hefei 230601,China)

机构地区：[1]安徽大学集成电路学院,合肥230601 [2]安徽省高性能集成电路工程研究中心,合肥230601

出　　处：《电子与封装》2024年第4期63-74,共12页Electronics & Packaging

摘　　要：忆阻器作为新原理器件,与现有互补金属氧化物半导体(CMOS)器件相比,具有结构简单、集成密度高、微缩性好、操作能耗低、易于三维集成等优点,是实现神经形态计算技术的理想存储器件之一。对近年来忆阻器基神经形态计算技术的研究进展进行综述,包括忆阻器基神经突触和神经元功能实现以及忆阻器基神经形态计算系统研究进展,并对当前忆阻器基神经形态计算技术仍然面临的挑战进行总结,对其前景做了展望。Memristor,as an emerging device with the advantages of simple structure,high integration density,excellent scaling property,low operating energy consumption,and easy three-dimensional integration property,compared with the existing complementary metal-oxide-semiconductor transistor(CMOS)devices,is one of the ideal memory devices for realizing neuromorphic computing technology.The research progress of memristor-based neuromorphic computing technology in recent years is reviewed.The main contents include the realization of memristor-based synapse and neuron functions,and the research of memristor-based neuromorphic computing system.The summary of the challenges faced by the memristor-based neuromorphic computing technology and the prospects are also given.

关 键 词：忆阻器 神经形态计算 突触 神经元 

分 类 号：TN389[电子电信—物理电子学]