面向神经形态系统的忆阻逻辑电路综述与展望

A Review and Perspectives on Memristor Logic Circuits for Neuromorphic Systems

神经形态系统泛指受脑启发、使用非冯诺依曼结构体系的新型信息处理系统,由于其高度的并行计算能力、极低的功耗和存算一体的特征,逐渐成为了人工智能领域的研究热点之一.其中,逻辑电路是神经形态计算硬件实现的基本计算单元,对于构建新型神经形态系统具有重要的意义.现有的应用于神经形态系统的逻辑电路大多由二极管、三极管等传统元器件或集成芯片组成,在尺寸、性能方面存在限制.具有纳米尺寸、电阻可变且非易失、耗能低等特性的忆阻器为构建面向神经形态系统中的逻辑电路提供了新的思路.基于此,本文对近年来的忆阻逻辑电路进行了详细的梳理和分析,基于电路中输入输出逻辑状态变量的不同,将其大致分为输入输出变量均为电阻(R-R型)电路、输入输出状态变量均为电压(V-V型)电路、输入输出状态变量为阻值和电压的组合(R-V型/V-R型)电路3类.同时,从组成部分、计算方式、逻辑状态变量、外围电路的复杂度、能量损耗、级联能力、延时性7个方面对3类忆阻逻辑电路进行全面的对比和分析,总结不同电路拓扑结构的忆阻逻辑电路的优缺点.进一步,整理分析了现有忆阻逻辑电路在联想记忆、情感计算、模式识别等领域的潜在应用.针对忆阻逻辑电路在操作复杂性和物理实现方面的缺陷,本文展望了忆阻逻辑电路的发展方向.最后,本文对忆阻逻辑电路及其在神经形态系统中的应用进行总结.

Neuromorphic system refers to a new type of information processing system inspired by brain and using non-Von Neumann structure system,which gradually becomes one of the research hotspots in the field of artificial intelligence due to its high parallel computing capability,extremely low power consumption and the characteristics of storage and computation integration.Among them,logic circuits are the basic computational units for neuromorphic computing hardware implementation,which are of great significance for building new neuromorphic systems.Most of the existing logic circuits for neuromorphic systems are composed of traditional components such as diodes and transistors or integrated chips,which have limitations in terms of size and performance.Memristors with nanometer size,variable and non-volatile resistance,and low energy consumption provide a new idea for building logic circuits for neuromorphic systems.Based on this,this paper presents a detailed review and analysis of recent memristor logic circuits,which are broadly classified into three categories based on the difference of input and output logic state variables in the circuits:circuits with both input and output variables as resistance(R-R type),circuits with both input and output state variables as voltage(V-V type),and circuits with a combination of resistance and voltage(R-V type/V-R type)as input and output state variables.The advantages and disadvantages of different circuit topologies of memristor logic circuits are summarized by comparing and analyzing the three types of memristor logic circuits in seven aspects:components,calculation methods,logic state variables,complexity of peripheral circuits,energy loss,cascading capability,and delay.Further,the potential applications of existing memristor logic circuits in the fields of associative memory,emotional computing,and pattern recognition are compiled and analyzed.In view of the shortcomings in operational complexity and physical implementation of memristor logic circuits,this paper foresees the development direction of memristor logic circuits.Finally,the paper concludes with a summary of the memristor logic circuits and their applications in neuromorphic systems.