Re-weighted Adversarial Adaptation Network for Unsupervised Domain Adaptation
Accepted version
Peer-reviewed
Repository URI
Repository DOI
Change log
Authors
Abstract
Unsupervised Domain Adaptation (UDA) aims to trans- fer domain knowledge from existing well-defined tasks to new ones where labels are unavailable. In the real-world applications, as the domain (task) discrepancies are usu- ally uncontrollable, it is significantly motivated to match the feature distributions even if the domain discrepancies are disparate. Additionally, as no label is available in the target domain, how to successfully adapt the classifier from the source to the target domain still remains an open ques- tion. In this paper, we propose the Re-weighted Adversarial Adaptation Network (RAAN) to reduce the feature distribu- tion divergence and adapt the classifier when domain dis- crepancies are disparate. Specifically, to alleviate the need of common supports in matching the feature distribution, we choose to minimize optimal transport (OT) based Earth- Mover (EM) distance and reformulate it to a minimax ob- jective function. Utilizing this, RAAN can be trained in an end-to-end and adversarial manner. To further adapt the classifier, we propose to match the label distribution and embed it into the adversarial training. Finally, after ex- tensive evaluation of our method using UDA datasets of varying difficulty, RAAN achieved the state-of-the-art re- sults and outperformed other methods by a large margin when the domain shifts are disparate.