摘要
本文主要论述了以专用信号处理芯片(DSP)为基础的、用于神经网络计算的多处理机系统,目的是利用神经网络方法实现信号的实时处理。本系统的特点是在充分考虑神经网络特性的基础上,构造了一个由四个处理单元组成的、具有共享多口存储器的多机系统。系统结构简单小巧,适于神经网络的运算。
This paper presents a DSP-based multi-processor system for artificial neural network (ANN) computing. The aim is to implement the real-time signal processing by ANN. The system, is characterized by a multiprocessor system with shared multiple-port RAM in view of the features of ANN. Its architecture is simple and is fit for computing of ANN.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1994年第2期157-160,共4页
Pattern Recognition and Artificial Intelligence
