论文部分内容阅读
相应于非线性系统用人工神经网络的逼近问题，本文引入了一种新的小波变换并研究了其性质.作为推论，本文给出了在Lp范数下单个隐层前馈神经网络逼近定理的构造性证明.
Corresponding to the approximation problem of artificial neural network for nonlinear systems, a new wavelet transform is introduced and its properties are studied.As a corollary, we present the approximation theorem for a single hidden layer feedforward neural network under Lp norm Constructive proof.