Approximation by series of sigmoidal functions with applications to neural networks (Q2255377): Difference between revisions

The authors develop a theory of uniform approximation of functions of certain classes by series of sigmoidal functions, i.e., functions \(\sigma\) on \({\mathbb R}\) such that \[ \lim_{x\to-\infty}\sigma(x)=0\quad\text{and}\quad\lim_{x\to\infty}\sigma(x)=1. \]