Unsupervised Domain Adaptation for Multitask Image Analysis in Realistic Context with Extreme Label Shift; Application to the CTAO first Large Sized Telescope
2026-08-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors study how to help a computer model trained on one set of labeled data work well on a related but different set of data that has no labels, a problem known as unsupervised domain adaptation. They focus on combining this with multitask learning when some classes are extremely rare, making the training more challenging. Their approach is tested on data from a physics experiment called the Cherenkov Telescope Array Observatory. They also explore ways to fix problems caused by big differences in class distribution between the data sets. The authors have shared their code and results openly for others to use.
Unsupervised domain adaptationMultitask learningClass imbalanceClass distribution shiftImportance weightingCherenkov Telescope Array ObservatoryExtreme label shiftAuxiliary tasksPhysics-based data
Authors
Michaël Dell'aiera, Thomas Vuillaume, Alexandre Benoit
Abstract
Unsupervised domain adaptation is a widespread set of methods that leverages the knowledge of a labeled source domain to train a model to perform well on a related unlabeled target domain. They generally introduce an auxiliary adaptation-related task that can be integrated into the multitask paradigm, which aims to merge multiple single-task models into a unified architecture. In this paper, we propose to associate domain adaptation and multitask balancing in the realistic context of an extreme class imbalance. Therefore, we propose a combined framework to cover and validate these approaches, and evaluate its performance in the physics-based context of the Cherenkov Telescope Array Observatory (CTAO). Along with a comparative study of some relevant adaptation techniques, we highlight the impact of extreme label shift and extend the investigations on importance weighting to rectify it. The complete code and results are published and available as open-source resources on Zenodo.