In this work, we propose an adversarial unsupervised domain adaptation (UDA) method under inherent conditional and label shifts, in which we aim to align the distributions w.r.t. both p(x|y) and p(y). Since labels are inaccessible in a target domain, conventional adversarial UDA methods assume that p(y) is invariant across domains and rely on aligning p(x) as an alternative to the p(x|y) alignment. To address this, we provide a thorough theoretical and empirical analysis of the conventional adversarial UDA methods under both conditional and label shifts, and propose a novel and practical alternative optimization scheme for adversarial UDA. Specifically, we infer the marginal p(y) and align p(x|y) iteratively at the training stage, and precisely align the posterior p(y|x) at the testing stage. Our experimental results demonstrate its effectiveness on both classification and segmentation UDA and partial UDA.
|Title of host publication||Proc. of ICCV'2021|
|Publication status||Published - Oct 2021|
|Event||2021 IEEE/CVF Int. Conf. on Computer Vision - Montreal, QC, Canada|
Duration: 10 Oct 2021 → 17 Oct 2021
|Conference||2021 IEEE/CVF Int. Conf. on Computer Vision|
|Period||10/10/21 → 17/10/21|