CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift
This addresses domain adaptation for machine learning models when label distributions differ between source and target domains, representing an incremental improvement over prior methods.
The paper tackles the problem of unsupervised domain adaptation under label distribution shift, where existing methods relying on covariate shift assumptions perform poorly, and proposes CASUAL to minimize conditional symmetric support divergence, achieving state-of-the-art results on benchmark tasks.
Unsupervised domain adaptation (UDA) refers to a domain adaptation framework in which a learning model is trained based on the labeled samples on the source domain and unlabeled ones in the target domain. The dominant existing methods in the field that rely on the classical covariate shift assumption to learn domain-invariant feature representation have yielded suboptimal performance under label distribution shift. In this paper, we propose a novel Conditional Adversarial SUpport ALignment (CASUAL) whose aim is to minimize the conditional symmetric support divergence between the source's and target domain's feature representation distributions, aiming at a more discriminative representation for the classification task. We also introduce a novel theoretical target risk bound, which justifies the merits of aligning the supports of conditional feature distributions compared to the existing marginal support alignment approach in the UDA settings. We then provide a complete training process for learning in which the objective optimization functions are precisely based on the proposed target risk bound. Our empirical results demonstrate that CASUAL outperforms other state-of-the-art methods on different UDA benchmark tasks under different label shift conditions.