Robust Correction of Sampling Bias using Cumulative Distribution Functions

Traditional machine learning approaches assume that training data and target data are drawn from the same distribution.