Domain Adaptation Using Pseudo Labels

TL;DR

This paper proposes a multi-stage pseudo-label refinement method using a pretrained network to improve unsupervised domain adaptation by accurately aligning categories between source and target domains.

Contribution

It introduces a simple, effective pseudo-label refinement procedure that enhances category alignment in unsupervised domain adaptation, outperforming complex state-of-the-art methods.

Findings

Improved target domain classification accuracy

Effective pseudo-label refinement based on confidence, distance, and consistency

Outperforms complex existing domain adaptation techniques

Abstract

In the absence of labeled target data, unsupervised domain adaptation approaches seek to align the marginal distributions of the source and target domains in order to train a classifier for the target. Unsupervised domain alignment procedures are category-agnostic and end up misaligning the categories. We address this problem by deploying a pretrained network to determine accurate labels for the target domain using a multi-stage pseudo-label refinement procedure. The filters are based on the confidence, distance (conformity), and consistency of the pseudo labels. Our results on multiple datasets demonstrate the effectiveness of our simple procedure in comparison with complex state-of-the-art techniques.