Comparison of rates of linear and neural network approximation
Věra Kůrková, Marcello Sanguineti · 2000
We develop some mathematical tools for comparison of rates of fixed versus variable basis function approximation. Using these tools, we describe sets of multivariable functions, for which lower bounds on worst-case errors in approximation by n-dimensional linear subspaces are larger than upper bounds on such errors in approximation by perceptron networks with n hidden units.