Dynamic Feature Alignment for Semi-supervised Domain Adaptation
This work addresses domain adaptation for scenarios where limited labeled target data is available, offering a practical solution with broad applicability in real-world machine learning tasks.
The paper tackles semi-supervised domain adaptation by proposing dynamic feature alignment to address inter- and intra-domain discrepancies, achieving significant improvements in state-of-the-art performance on datasets like DomainNet and Office-Home.
Most research on domain adaptation has focused on the purely unsupervised setting, where no labeled examples in the target domain are available. However, in many real-world scenarios, a small amount of labeled target data is available and can be used to improve adaptation. We address this semi-supervised setting and propose to use dynamic feature alignment to address both inter- and intra-domain discrepancy. Unlike previous approaches, which attempt to align source and target features within a mini-batch, we propose to align the target features to a set of dynamically updated class prototypes, which we use both for minimizing divergence and pseudo-labeling. By updating based on class prototypes, we avoid problems that arise in previous approaches due to class imbalances. Our approach, which doesn't require extensive tuning or adversarial training, significantly improves the state of the art for semi-supervised domain adaptation. We provide a quantitative evaluation on two standard datasets, DomainNet and Office-Home, and performance analysis.