研究多元插值神经网络的收敛性问题,举例说明多元插值神经网络对连续函数并非一致收敛,回答了谢庭藩等所提出的问题.
This paper deals with the convergence of multidimensional interpolation neural networks. It is shown by a constructive example that the multidimensional interpolation neural networks are not sure to converge uniformly for any multivariate continuous function defined by a compact set. And a problem proposed by Xie Tingfan et al is answered.