Learning a data representation that can be effectively transferred to downstream supervised learning tasks in an unlabeled scenario is both critical and challenging. In this talk, we propose novel app ...