Key points are not available for this paper at this time.
The most basic assumption used in statistical learning theory is that training data and test data are drawn from the same underlying distribution. Unfortunately, in many applications, the in-domai...
DauméHal et al. (Mon,) studied this question.