关于周期神经网络逼近阶的研究(英文)  

On the Investigation of the Degree of Approximation by Periodic Neural Networks

作　　者：盛宝怀[1] 周观珍[2] 刘三阳[3] 

机构地区：[1]绍兴文理学院数学系 [2]宁波大学数学系 [3]西安电子科技大学应用数学系

出　　处：《运筹学学报》2005年第4期21-30,共10页Operations Research Transactions

基　　金：the National Natural Science Foundation of China(10371024)Foundation of Zhejiang Education Committee(No. 20030431)the Natural Science Foundation of Zhejiang Province(Y604003)the Doctor Foundation of Ningbo City(No. 2004A620017).

摘　　要：借助于有关Fourier级数的Riesz平均构造出了一类含有一个隐含层的周期神经网络与平移网络,与已有的讨论相比较,在获得相同的逼近阶的情况下,此类网络的隐层单元要求较少的神经元个数．A kind of periodic neural and translation networks is respectively constructed by the Riesz means of the Fourier series. Compared with the same kind of known neural and translation networks, the networks of this paper have the advantages of less neurons and trans, though the same degree of approximation can be obtained.

关 键 词：运筹学 周期神经网络 逼近阶 神经元个数 

分 类 号：O174.41[理学—数学] TP18[理学—基础数学]