Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift
复制标题
DOI：
--
复制
发表时间：
2020-03
期刊：
影响因子：
--
通讯作者：
Rémi Tachet des Combes;Han Zhao;Yu-Xiang Wang;Geoffrey J. Gordon
中科院分区：
文献类型：
--
作者：
Rémi Tachet des Combes;Han Zhao;Yu-Xiang Wang;Geoffrey J. Gordon
对抗学习在无监督域自适应设置中表现出良好的性能，通过学习在源域上表现良好的域不变表示。然而，最近的工作强调了现有方法在源和靶域之间存在不匹配的标签分布的限制。在本文中，我们将最近的对抗域自适应性能上限扩展到多类分类和更一般的鉴别器。然后，我们提出广义标签移位（GLS）作为一种方法来提高对不匹配的标签分布的鲁棒性。GLS指出，在标签的条件下，存在一个在源域和目标域之间不变的输入表示。在GLS下，我们对任何分类器的传输性能提供理论保证。我们还设计了GLS成立的充分必要条件。这些条件是基于对域之间的相对类权重的估计和对样本的适当重新加权。在我们的理论见解的指导下，我们修改了三个广泛使用的算法，JAN，DANN和CDAN，并评估了它们在标准域自适应任务中的性能，其中我们的方法优于基础版本。我们还展示了显着的收益，人工创建的任务，其源和目标标签分布之间的巨大分歧。
Adversarial learning has demonstrated good performance in the unsupervised domain adaptation setting, by learning domain-invariant representations that perform well on the source domain. However, recent work has underlined limitations of existing methods in the presence of mismatched label distributions between the source and target domains. In this paper, we extend a recent upper-bound on the performance of adversarial domain adaptation to multi-class classification and more general discriminators. We then propose generalized label shift (GLS) as a way to improve robustness against mismatched label distributions. GLS states that, conditioned on the label, there exists a representation of the input that is invariant between the source and target domains. Under GLS, we provide theoretical guarantees on the transfer performance of any classifier. We also devise necessary and sufficient conditions for GLS to hold. The conditions are based on the estimation of the relative class weights between domains and on an appropriate reweighting of samples. Guided by our theoretical insights, we modify three widely used algorithms, JAN, DANN and CDAN and evaluate their performance on standard domain adaptation tasks where our method outperforms the base versions. We also demonstrate significant gains on artificially created tasks with large divergences between their source and target label distributions.