Unsupervised domain adaptation adapts classifiers to an un-labeled target domain by exploiting a labeled source domain. To reduce discrepancy between source and target domains, adversarial learning ...