GMM-COMET: Continual Source-Free Universal Domain Adaptation via a Mean Teacher and Gaussian Mixture Model-Based Pseudo-Labeling
Pascal Schlachter, Bin Yang
TL;DR
GMM-COMET tackles continual source-free universal domain adaptation by unifying a Gaussian Mixture Model–based pseudo-labeling scheme with a mean-teacher online adaptation framework and additional consistency regularization. The method operates on a stream of unlabeled target domains, using a reduced-feature GMM to assign pseudo-labels for known source classes, while detecting unknowns via adaptive OOD thresholds. It integrates a contrastive loss and an entropy-based loss, plus two consistency losses (source and student–teacher) to stabilize long-horizon adaptation, optimizing a joint objective. Empirically, GMM-COMET consistently outperforms the source-only baseline across diverse domain shifts and datasets, establishing a strong inaugural benchmark for continual SF-UniDA and providing a robust, practical baseline for future research.
Abstract
Unsupervised domain adaptation tackles the problem that domain shifts between training and test data impair the performance of neural networks in many real-world applications. Thereby, in realistic scenarios, the source data may no longer be available during adaptation, and the label space of the target domain may differ from the source label space. This setting, known as source-free universal domain adaptation (SF-UniDA), has recently gained attention, but all existing approaches only assume a single domain shift from source to target. In this work, we present the first study on continual SF-UniDA, where the model must adapt sequentially to a stream of multiple different unlabeled target domains. Building upon our previous methods for online SF-UniDA, we combine their key ideas by integrating Gaussian mixture model-based pseudo-labeling within a mean teacher framework for improved stability over long adaptation sequences. Additionally, we introduce consistency losses for further robustness. The resulting method GMM-COMET provides a strong first baseline for continual SF-UniDA and is the only approach in our experiments to consistently improve upon the source-only model across all evaluated scenarios. Our code is available at https://github.com/pascalschlachter/GMM-COMET.