Both class imbalance and covariate shift simplify the difference between and . It is still possible that differs from or differs from .
Jiang and Zhai (2007a) considered the case when differs from , and proposed a heuristic method to remove ``misleading'' training instances from the source domain, where is very different from . To discover these ``misleading'' training instances, some labeled data from the target domain is needed. This method therefore is only suitable for supervised domain adaptation.