Simultaneous approximations of multivariate functions and their derivatives by neural networks with one hidden layer (Q2564066)

Let \(\sigma\) be a non-polynomial activation function of a neural network that has \(n\)th order continuous derivatives on \(R\). This paper shows that for any compact set \(K\) of \(R^s\), \(s\geq 1\), and any multivariate function \(f\) defined on an open set containing \(K\), a neural network with one hidden layer can be so constructed that \(f\)and all its existing continuous \(k\)th order partial derivatives, for \(k=(k_1,\dots, k_s)\in \mathbb{Z}^s_+\) satisfying \(\sum^s_{i=1} k_i\leq n\), can be simultaneously and uniformly approximated on \(K\) by the network.