Weakly Supervised Open-set Domain Adaptation by Dual-domain Collaboration
This work addresses a practical scenario in domain adaptation for machine learning applications where both domains are partially labeled and have different classes, which is incremental but relevant for real-world tasks like person reidentification.
The paper tackles the problem of weakly supervised open-set domain adaptation, where both source and target domains are partially labeled and have non-overlapping classes, by proposing the Collaborative Distribution Alignment (CDA) method to enable bilateral knowledge transfer for classifying unlabeled data and identifying outliers, achieving state-of-the-art performance on the Office benchmark and in person reidentification applications.
In conventional domain adaptation, a critical assumption is that there exists a fully labeled domain (source) that contains the same label space as another unlabeled or scarcely labeled domain (target). However, in the real world, there often exist application scenarios in which both domains are partially labeled and not all classes are shared between these two domains. Thus, it is meaningful to let partially labeled domains learn from each other to classify all the unlabeled samples in each domain under an open-set setting. We consider this problem as weakly supervised open-set domain adaptation. To address this practical setting, we propose the Collaborative Distribution Alignment (CDA) method, which performs knowledge transfer bilaterally and works collaboratively to classify unlabeled data and identify outlier samples. Extensive experiments on the Office benchmark and an application on person reidentification show that our method achieves state-of-the-art performance.