Abstract

We propose interpolating nonparametric regression estimators that attain optimal convergence rates adaptively to the smoothness.