线性和型人工神经网络的拓扑结构和统一描述

Topological Structure and Unified Representation of Linear Additive Artificial Neural Networks

本文从控制系统的角度讨论了线性和型人工神经网络的拓扑结构和统一描述问题,对一些典型网络模型的分析表明:采用控制系统理论中关于系统描述与结构分析的传递矩阵和框图方法来分析神经网络,不仅可使网络的结构清晰化,而且还可为网络的综合提供一条系统的思路.

This paper deals with topological structures and unified representation of linear additive artificial neural networks from the view point of control systems. An analysis of some typical model neural networks shows that, by the block diagram approach for structures and descriptions used in control system theory, the topology of neural networks is clarified and network synthesis is tractable.