模糊ARTMAP神经网络在二维信息聚类分析中的应用  

THE FUZZY ARTMAP NEURAL NETWORK'S APPLICATION IN ANALYZING TWO-DIMENSIONAL SIGNALS

作　　者：秦安文[1] 

机构地区：[1]山东大学电力工程学院,济南250061

出　　处：《山东工业大学学报》2000年第5期450-455,461,共7页

摘　　要：在模糊 ARTMAP神经网络的模型的基础上 ,重点介绍了二维信息的聚类原理 ;并从几何的角度 ,讨论了二维信息的类描述法 ,类的物理含义、类的扩充及界定等问题 .同时 ,对类的唯一性、收敛性和稳定性在理论上进行了探讨 .最后以圆的识别问题作为模拟教师样本对网络进行了学习 ,给出了学习步骤及结果 .该方法特别适合于分析那些无确定的数学模型的事件 .Based on the architecture of fuzzy ARTMAP neural network, the paper introduces cluster analysis for two/dimensional signals, from the point of plane geonetry discusses the classifical physical definition, expandsion, and limitation and makes a theoretical study on classifications in terms of the simpleness, contraction and stability. At the end, the paper teaches the architecture by circle recognition and provides the steps and results for the architecture learning.It is very useful for the discussed method to analyze those events which can not be done by a definitive mathmatic architecture.

关 键 词：神经元模型 类分离程度 激励 模糊 ARTMAP神经网络 二维信息 聚类分析 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]