与神经网络有关的逼近问题

APPROXIMATION PROBLEM CONCERNED WITH NEURAL NETWORKS

本文来自神经网络表示能力问题的研究，主要讨论单个函数满足什么条件其所有伸缩和平移的线性组合在某类函数空间中稠密．本文结果对于神经网络理论研究具有重要意义．

This paper results from the research work of representation capability of neural networks,which discusses what kinds of conditions are imposed on single function such that it's linear combinations with it's translations and dilations are dense in some function space . The results of this paper are of great importance in the study of neural networks theory.