Your Classifier can Secretly Suffice Multi-Source Domain Adaptation

TL;DR

This paper introduces a novel approach to multi-source domain adaptation that leverages implicit domain alignment through classifier agreement on pseudo-labeled target data, eliminating the need for explicit distribution alignment objectives.

Contribution

The work proposes Self-supervised Implicit Alignment (SImpAl), a method that uses classifier agreement on pseudo-labels for domain adaptation without additional training objectives.

Findings

Effective even under category-shift among source domains.

Classifier agreement can reliably indicate training convergence.

Achieves competitive results on five benchmark datasets.

Abstract

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.