Confidence Score for Source-Free Unsupervised Domain Adaptation

TL;DR

This paper introduces a novel confidence score, JMDS, that leverages both source and target domain knowledge for source-free unsupervised domain adaptation, improving sample importance estimation and achieving state-of-the-art results.

Contribution

The paper proposes the JMDS confidence score and the CoWA-JMDS framework, combining sample-wise importance weighting with a new Mixup variant for better domain adaptation.

Findings

JMDS outperforms existing confidence scores.

CoWA-JMDS achieves state-of-the-art performance.

Effective in various SFUDA scenarios.

Abstract

Source-free unsupervised domain adaptation (SFUDA) aims to obtain high performance in the unlabeled target domain using the pre-trained source model, not the source data. Existing SFUDA methods assign the same importance to all target samples, which is vulnerable to incorrect pseudo-labels. To differentiate between sample importance, in this study, we propose a novel sample-wise confidence score, the Joint Model-Data Structure (JMDS) score for SFUDA. Unlike existing confidence scores that use only one of the source or target domain knowledge, the JMDS score uses both knowledge. We then propose a Confidence score Weighting Adaptation using the JMDS (CoWA-JMDS) framework for SFUDA. CoWA-JMDS consists of the JMDS scores as sample weights and weight Mixup that is our proposed variant of Mixup. Weight Mixup promotes the model make more use of the target domain knowledge. The experimental results show that the JMDS score outperforms the existing confidence scores. Moreover, CoWA-JMDS achieves state-of-the-art performance on various SFUDA scenarios: closed, open, and partial-set scenarios.