Learning Purified Feature Representations from Task-irrelevant Labels

Yinghui Li, Chen Wang, Yangning Li +2

Learning an empirically effective model with generalization using limited data is a challenging task for deep neural networks. In this paper, we propose a novel learning framework…