@inproceedings{aff24825cae44615aeca0403b454d5e0,
title = "The essential approximation order for neural networks with trigonometric hidden layer units",
abstract = "There have been various studies on approximation ability of feedforward neural networks. The existing studies are, however, only concerned with the density or upper bound estimation on how a multivariate function can be approximated by the networks, and consequently, the essential approximation ability of networks cannot be revealed. In this paper, by establishing both upper and lower bound estimations on approximation order, the essential approximation ability of a class of feedforward neural networks with trigonometric hidden layer units is clarified in terms of the second order modulus of smoothness of approximated function.",
author = "Chunmei Ding and Feilong Cao and Zongben Xu",
year = "2006",
doi = "10.1007/11759966\_11",
language = "英语",
isbn = "354034439X",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "72--79",
booktitle = "Advances in Neural Networks - ISNN 2006",
note = "3rd International Symposium on Neural Networks, ISNN 2006 - Advances in Neural Networks ; Conference date: 28-05-2006 Through 01-06-2006",
}