17
218

Does data interpolation contradict statistical optimality?

Abstract

We show that learning methods interpolating the training data can achieve optimal rates for the problems of nonparametric regression and prediction with square loss.

View on arXiv
Comments on this paper