Your Classifier can Secretly Suffice Multi-Source Domain Adaptation
This work addresses domain adaptation for machine learning practitioners by offering a simpler, more efficient approach that reduces training complexity, though it is incremental as it builds on existing implicit alignment observations.
The paper tackles multi-source domain adaptation by showing that deep models can implicitly align domains under label supervision, eliminating the need for auxiliary alignment objectives. The proposed Self-supervised Implicit Alignment (SImpAl) method achieves competitive results on five benchmarks, even under category-shift among source domains.
Multi-Source Domain Adaptation (MSDA) deals with the transfer of task knowledge from multiple labeled source domains to an unlabeled target domain, under a domain-shift. Existing methods aim to minimize this domain-shift using auxiliary distribution alignment objectives. In this work, we present a different perspective to MSDA wherein deep models are observed to implicitly align the domains under label supervision. Thus, we aim to utilize implicit alignment without additional training objectives to perform adaptation. To this end, we use pseudo-labeled target samples and enforce a classifier agreement on the pseudo-labels, a process called Self-supervised Implicit Alignment (SImpAl). We find that SImpAl readily works even under category-shift among the source domains. Further, we propose classifier agreement as a cue to determine the training convergence, resulting in a simple training algorithm. We provide a thorough evaluation of our approach on five benchmarks, along with detailed insights into each component of our approach.