Near-optimal learning with average H\"older smoothness

Open in new window