# Radio Galaxy Zoo: Using semi-supervised learning to leverage large unlabelled data-sets for radio galaxy classification under data-set shift

arXiv: https://arxiv.org/abs/2204.08816 · HTML: https://mlsanity.com/paper/2204.08816/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Builds on 0 of its 15 references that are in the map.
- First posted on arXiv 19 Apr 2022.

## Built on by (most cited first, 8 of 8)

- [MiraBest: A Dataset of Morphologically Classified Radio Galaxies for Machine Learning](https://mlsanity.com/paper/2305.11108/) (2023)
- [Evaluating Bayesian deep learning for radio galaxy classification](https://mlsanity.com/paper/2405.18351/) (2024)
- [A review of unsupervised learning in astronomy](https://mlsanity.com/paper/2406.17316/) (2024)
- [Morphological Classification of Radio Galaxies using Semi-Supervised Group Equivariant CNNs](https://mlsanity.com/paper/2306.00031/) (2023)
- [Advances on the classification of radio image cubes](https://mlsanity.com/paper/2305.03435/) (2023)
- [Deep Learning for Morphological Identification of Extended Radio Galaxies using Weak Labels](https://mlsanity.com/paper/2308.05166/) (2023)
- [Can AI Dream of Unseen Galaxies? Conditional Diffusion Model for Galaxy Morphology Augmentation](https://mlsanity.com/paper/2506.16233/) (2025)
- [Spatio-Spectroscopic Representation Learning using Unsupervised Convolutional Long-Short Term Memory Networks](https://mlsanity.com/paper/2602.18426/) (2026)

## Abstract (arXiv)

In this work we examine the classification accuracy and robustness of a state-of-the-art semi-supervised learning (SSL) algorithm applied to the morphological classification of radio galaxies. We test if SSL with fewer labels can achieve test accuracies comparable to the supervised state-of-the-art and whether this holds when incorporating previously unseen data. We find that for the radio galaxy classification problem considered, SSL provides additional regularisation and outperforms the baseline test accuracy. However, in contrast to model performance metrics reported on computer science benchmarking data-sets, we find that improvement is limited to a narrow range of label volumes, with performance falling off rapidly at low label volumes. Additionally, we show that SSL does not improve model calibration, regardless of whether classification is improved. Moreover, we find that when different underlying catalogues drawn from the same radio survey are used to provide the labelled and unlabelled data-sets required for SSL, a significant drop in classification performance is observered, highlighting the difficulty of applying SSL techniques under dataset shift. We show that a class-imbalanced unlabelled data pool negatively affects performance through prior probability shift, which we suggest may explain this performance drop, and that using the Frechet Distance between labelled and unlabelled data-sets as a measure of data-set shift can provide a prediction of model performance, but that for typical radio galaxy data-sets with labelled sample volumes of O(1000), the sample variance associated with this technique is high and the technique is in general not sufficiently robust to replace a train-test cycle.