In this work, we propose an adversarial unsupervised domain adaptation (UDA) approach with the inherent conditional and label shifts, in which we aim to align the distributions w.r.t. both $p(x|y)$ and $p(y)$. Since the label is inaccessible in the target domain, the conventional adversarial UDA assumes $p(y)$ is invariant across domains, and relies 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 in the training, and precisely align the posterior $p(y|x)$ in testing. Our experimental results demonstrate its effectiveness on both classification and segmentation UDA, and partial UDA.

中文翻译：

---

具有条件和标签移位的对抗性无监督域适应：推断、对齐和迭代

在这项工作中，我们提出了一种具有固有条件和标签转换的对抗性无监督域适应 (UDA) 方法，我们的目标是对齐 $p(x|y)$ 和 $p(y)$ 的分布。由于标签在目标域中是不可访问的，传统的对抗性 UDA 假设 $p(y)$ 跨域是不变的，并依赖于对齐 $p(x)$ 作为 $p(x|y)$ 对齐的替代. 为了解决这个问题，我们对条件和标签转换下的传统对抗性 UDA 方法进行了彻底的理论和实证分析，并提出了一种新颖实用的对抗性 UDA 替代优化方案。具体来说，我们在训练中迭代地推断出边际 $p(y)$ 并对齐 $p(x|y)$，并在测试中精确对齐后验 $p(y|x)$。