On Label Shift in Domain Adaptation via Wasserstein Distance
This addresses domain adaptation challenges for machine learning applications where label distributions differ between domains, but it is incremental as it builds on existing optimal transport methods.
The paper tackles the label shift problem in domain adaptation by using optimal transport to align source and target domains, proposing LDROT to mitigate both data and label shifts, and shows competitive results in experiments.
We study the label shift problem between the source and target domains in general domain adaptation (DA) settings. We consider transformations transporting the target to source domains, which enable us to align the source and target examples. Through those transformations, we define the label shift between two domains via optimal transport and develop theory to investigate the properties of DA under various DA settings (e.g., closed-set, partial-set, open-set, and universal settings). Inspired from the developed theory, we propose Label and Data Shift Reduction via Optimal Transport (LDROT) which can mitigate the data and label shifts simultaneously. Finally, we conduct comprehensive experiments to verify our theoretical findings and compare LDROT with state-of-the-art baselines.