Random Rates in Anisotropic Regression : Discussion