Unsupervised domain adaptation for multitask image analysis in realistic context with extreme label shift; Application to the CTAO first large sized telescope
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 Dell’aiera et al. (2024). TESTING VERSIONING
Related
- Self-calibration of adaptive optics systems. Application to the THEMIS solar telescope
- Unsupervised selection and characterisation of little red dots in JWST surveys with manifold learning
- Planetesimal Formation under Realistic Gas Dynamics: Dust Concentration in Turbulent and Windy Protoplanetary Disks
- Planetesimal Formation under Realistic Gas Dynamics: Dust Concentration in Turbulent and Windy Protoplanetary Disks
- CtaAgnVar : a Pipeline for Variability Studies with the Cherenkov Telescope Array Observatory
BibTeX
@article{dellaiera2026unsupervised,
title = {{Unsupervised domain adaptation for multitask image analysis in realistic context with extreme label shift; Application to the CTAO first large sized telescope}},
author = {Michael Dell'Aiera and Thomas Vuillaume and Alexandre Benoit},
journal = {Astron.Comput.},
year = {2026},
month = sep,
doi = {10.1016/j.ascom.2026.101176},
url = {https://hal.science/hal-05733765v1},
note = {CC BY},
}
RIS
TY - JOUR
TI - Unsupervised domain adaptation for multitask image analysis in realistic context with extreme label shift; Application to the CTAO first large sized telescope
AU - Michael Dell'Aiera
AU - Thomas Vuillaume
AU - Alexandre Benoit
PY - 2026
DA - 2026/09/01
JO - Astron.Comput.
DO - 10.1016/j.ascom.2026.101176
UR - https://hal.science/hal-05733765v1
AB - 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 Dell’aiera et al. (2024). TESTING VERSIONING
ER -