Subtype-aware Unsupervised Domain Adaptation for Medical Diagnosis
This work aims to improve the accuracy of medical diagnosis for patients by accounting for subtle, underlying disease subtypes in UDA settings, which is an incremental improvement to existing UDA methods.
This paper addresses the challenge of unsupervised domain adaptation (UDA) in medical diagnosis by focusing on fine-grained subtype structures within classes. They propose a method that enforces class-wise separation and subtype-wise compactness using intermediate pseudo labels, achieving promising results on medical diagnosis tasks.
Recent advances in unsupervised domain adaptation (UDA) show that transferable prototypical learning presents a powerful means for class conditional alignment, which encourages the closeness of cross-domain class centroids. However, the cross-domain inner-class compactness and the underlying fine-grained subtype structure remained largely underexplored. In this work, we propose to adaptively carry out the fine-grained subtype-aware alignment by explicitly enforcing the class-wise separation and subtype-wise compactness with intermediate pseudo labels. Our key insight is that the unlabeled subtypes of a class can be divergent to one another with different conditional and label shifts, while inheriting the local proximity within a subtype. The cases of with or without the prior information on subtype numbers are investigated to discover the underlying subtype structure in an online fashion. The proposed subtype-aware dynamic UDA achieves promising results on medical diagnosis tasks.