B_a空间中神经网络和平移网络的逼近

摘要/Abstract

摘要：

讨论了具一个隐层单元的神经网络在B_a空间中逼近的特征性定理并给出了逼近估计.对于平移网络,建立了Favard型估计.Orlicz空间中的相应结果均作为应用而给出.

Abstract:

The characteristic theorem as well as the degree of approximating by neural ne tworks and translation networks, with a single hidden layer in B_a spaces is established. The Favard type estimate is obtained as well. The correspondences of Orlicz spaces are given by one of the applications of B_a spaces.


Key words: Neural networks; Translation networks;B_a spaces, Approximation.

中图分类号:

41A20

周观珍. B_a空间中神经网络和平移网络的逼近[J]. 数学物理学报, 2005, 25(4): 569-576.

ZHOU Guan-Zhen. On Approximation by Neural and Translation Networks in B_a Spaces[J]. Acta mathematica scientia,Series A, 2005, 25(4): 569-576.

参考文献

[1]Ding X Q, Luo P Z. B_a spaces and some estimates of Laplace operators. J Sys Sci Math Sci, 1981, 1(1):9-33

[2]Chen G R, Men B Q. Interpolation of B_a spaces. Acta Math Scientia, 19 88, 8(1): 65-70

[3]Ding X Q, He Y Z, Luo P Z. Theory of B_a Space and its Application. Beijing: Science Press, 1992

[4]盛保怀.B_a空间中的算子逼近. 数学物理学报, 1991,11( 4):387-395

[5]盛保怀B_a空间中Kantorovich算子的饱和性. 数学杂志, 1992,12(2):146-154

[6]盛宝怀,刘三阳, 陈广荣. Kantorovi算子在B_a Besov空间中的饱和性. 应用数学学报, 2002,25(3):392-401

[7]Sheng B H, Liu S Y, Chen G R. The MarcinkiewiczZygmund inequality in B_a spaces(I). J Sys Sci and Compl, 2003,16(1):74-84

[8]Chen T P, Chen H, Lin R W. Approximation capablity in C(R^n) by multilayer feed forward networks and related problems. IEEE Tran Neural Networks, 1995,6(1):25-30

[9]Chen T P, Chen H. Approximation capablity to functions of severa l variables, nonlinear functions, and operators by radial basis function neural networks. IEEE Tran Neural Networks, 1995, 6(4):904-910

[10]Chen T P, Chen H. Universal approximation to nonlinear operators by neural networks with arbitrary action functions and its application to dynamical system. IEEE Tran Neural Network s, 1995,6(4):911-917

[11]Chen T P. A unified approach for neural networklike approximatio n of nonlinear functionals. Neural Networks, 1998,11(5):981-983

[12]Mhaskar H N, Micchelli C A. Degree of approximation by neural and translation networks with single hidden layer. Advan in Appli Math,1995,16(1): 151-183

[13]蒋传海, 吴和兵, 李纯明. 多元周期函数的一类逼近及逼近阶估计. 数学学报, 1 999, 42(2): 263-270

[14]王建力, 盛宝怀, 周颂平. 非周期神经网络及平移网络在L\+p\-w中的逼近. 数学学报, 2003, 46(1): 65-74

[15]周观珍，盛宝怀。一类球面带形平移网络算子的逼近。数学物理学报，2005，25A(2)：269-276

[16]吴从,王庭辅. 奥尔里奇空间及其应用. 哈尔滨: 哈尔滨科学技术出版社,1983

[17]Deng Y H, Gu Y G. A new class of function spaces and the Hilbert transfo rm. J Math Anal Appl, 1985, 108(1): 99-106