domain adaptation aims at adapting a prediction function trained on a source domain, for a new different but related target domain. Recently several subspace learning methods have proposed adaptive solutions in t