Exponential Convergence of the Deep Neural Network Approximation for Analytic Functions

Abstract

We prove that for analytic functions in low dimension, the convergence rate of the deep neural network approximation is exponential.