摘要
本文讨论了正则模糊神经网络对于定义于区间［０，Ｔ０］上连续可减模糊值函数的普遍近似性，在此基础上，证明了［０，Ｔ０］上取值为三角形模糊数的连续递增函数可用正则模糊神经网络逼近到任意精度．
In this paper, it is argured that regular fuzzy neural networks can be universal approximators to continuously subtractable fuzzy valued functions, defined on the interval [0, T0]. Then arbitrarily accurate approximations of triangle fuzzy number valued functions that are continuously increasin on [0, T0] by regularfuzzy neural networks are showed.
出处
《电子学报》
EI
CAS
CSCD
北大核心
1997年第11期41-45,共5页
Acta Electronica Sinica
基金
国防预研基金
关键词
正则模糊
神经网络
可减模糊值函数
Regular fuzzy neural network, Subtractable fuzzy valued function, Sigmoidal function, Triangle fuzzy number
