摘要
论文研究了一种新型联想记忆神经网络的硬件实现,并给出了运行框图及其说明。每个神经元由一个功能简单的微处理单元和一组存储单元构成,存储在连接中的数据,转移到相应的目标神经元的E2PROM中,由一个微处理器控制所有神经元的运行和作为与外部设备的接口,硬件网络本身同时实现了回忆和记忆功能。
This paper presents a method of hardware implementation on a new a ss ociative memory neural network,pro-vides the working block diagram and explain s it in detail.Each neuron is composed of a MPU with simple structure and a set of memory unit,the data stored in links are moved to E 2 PROM of target neurons,one MPU controls the oper-ation of all neuron and w ork as a interface between the network and outside device,the hardware network itself can do both memorizing and recalling.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第14期25-29,共5页
Computer Engineering and Applications
基金
广东省自然科学基金项目(编号:011626)
广东省科技厅工业攻关项目(编号:c10909)
关键词
神经网络
联想记忆
Hamming距
硬件实现
neural network,associat ive memory,Hamming distance,Hardware implementation