Enabling Unsupervised Discovery in Astronomical Images through Self-Supervised Representations

Koketso Mohale; Michelle Lochner

RECORD

Enabling Unsupervised Discovery in Astronomical Images through Self-Supervised Representations

Koketso Mohale, Michelle Lochner

Instrumentation and Methods for Astrophysics · Submitted 23 Nov 2023

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim This work aims to demonstrate that applying deep representation learning is key to unlocking the potential of unsupervised discovery in future datasets from telescopes such as the Vera C.

how it is stated · asserted 0.90 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

arXiv ID2311.14157 (version 2)

Pith IDCWBVULCK, Pith's permanent record of this paper and its review

CitationKoketso Mohale, Michelle Lochner. "Enabling Unsupervised Discovery in Astronomical Images through Self-Supervised Representations." arXiv:2311.14157 (2023). https://pith.science/paper/CWBVULCK

Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 8,143 input tokens.

What the abstract itself states, as probabilities: claims first 0.03 · states limitation 0.69 · human participants 0.02 · controlled comparison 0.07 · randomized assignment 0.01 · states quantitative result 0.04 · compares with prior results 0.11 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v2. Results and the main claim are underlined; show every role.

Unsupervised learning, a branch of machine learning that can operate on unlabelled data, has proven to be a powerful tool for data exploration and discovery in astronomy. As large surveys and new telescopes drive a rapid increase in data size and richness, these techniques offer the promise of discovering new classes of objects and of efficient sorting of data into similar types. However, unsupervised learning techniques generally require feature extraction to derive simple but informative representations of images. In this paper, we explore the use of self-supervised deep learning as a method of automated representation learning. We apply the algorithm Bootstrap Your Own Latent (BYOL) to Galaxy Zoo DECaLS images to obtain a lower dimensional representation of each galaxy, known as features. We briefly validate these features using a small supervised classification problem. We then move on to apply an automated clustering algorithm, demonstrating that this fully unsupervised approach is able to successfully group together galaxies with similar morphology. The same features prove useful for anomaly detection, where we use the framework astronomaly to search for merger candidates. While the focus of this work is on optical images, we also explore the versatility of this technique by applying the exact same approach to a small radio galaxy dataset. This work aims to demonstrate that applying deep representation learning is key to unlocking the potential of unsupervised discovery in future datasets from telescopes such as the Vera C. Rubin Observatory and the Square Kilometre Array.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#	sentence	role	the authors' own work	states the main conclusion	strength
s0	Unsupervised learning, a branch of machine learning that can operate on unlabelled data, has proven to be a powerful tool for data exploration and discovery in astronomy.	background 1.00	0.05	0.04	asserted 1.00
s1	As large surveys and new telescopes drive a rapid increase in data size and richness, these techniques offer the promise of discovering new classes of objects and of efficient sorting of data into similar types.	background 1.00	0.05	0.04	hedged possibility 0.96
s2	However, unsupervised learning techniques generally require feature extraction to derive simple but informative representations of images.	gap 0.51	0.04	0.02	asserted 0.97
s3	In this paper, we explore the use of self-supervised deep learning as a method of automated representation learning.	contribution 0.82	0.79	0.05	asserted 0.64
s4	We apply the algorithm Bootstrap Your Own Latent (BYOL) to Galaxy Zoo DECaLS images to obtain a lower dimensional representation of each galaxy, known as features.	method 0.98	0.97	0.04	asserted 0.97
s5	We briefly validate these features using a small supervised classification problem.	method 0.85	0.96	0.03	asserted 0.75
s6	We then move on to apply an automated clustering algorithm, demonstrating that this fully unsupervised approach is able to successfully group together galaxies with similar morphology.	result 0.82	0.96	0.27	asserted 1.00
s7	The same features prove useful for anomaly detection, where we use the framework astronomaly to search for merger candidates.	contribution 0.46	0.94	0.07	asserted 1.00
s8	While the focus of this work is on optical images, we also explore the versatility of this technique by applying the exact same approach to a small radio galaxy dataset.	contribution 0.73	0.95	0.05	asserted 0.83
s9	This work aims to demonstrate that applying deep representation learning is key to unlocking the potential of unsupervised discovery in future datasets from telescopes such as the Vera C.	future work 0.50	0.23	0.62	asserted 0.90
s10	Rubin Observatory and the Square Kilometre Array.	future work 0.85	0.10	0.12	unclear 0.91

What kind of paper the classifier reports this is: method or tool 0.79 · evidence from analysis of existing data 0.90.

What was examined, and what was not

sources read

arXiv listing abstract v2

not examined

full text
figures and tables
references and citing sentences
relations to other papers

levels run

P abstract

not run

S sections
T body sentences
C citations
R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper

Original abstract

Unsupervised learning, a branch of machine learning that can operate on unlabelled data, has proven to be a powerful tool for data exploration and discovery in astronomy. As large surveys and new telescopes drive a rapid increase in data size and richness, these techniques offer the promise of discovering new classes of objects and of efficient sorting of data into similar types. However, unsupervised learning techniques generally require feature extraction to derive simple but informative representations of images. In this paper, we explore the use of self-supervised deep learning as a method of automated representation learning. We apply the algorithm Bootstrap Your Own Latent (BYOL) to Galaxy Zoo DECaLS images to obtain a lower dimensional representation of each galaxy, known as features. We briefly validate these features using a small supervised classification problem. We then move on to apply an automated clustering algorithm, demonstrating that this fully unsupervised approach is able to successfully group together galaxies with similar morphology. The same features prove useful for anomaly detection, where we use the framework astronomaly to search for merger candidates. While the focus of this work is on optical images, we also explore the versatility of this technique by applying the exact same approach to a small radio galaxy dataset. This work aims to demonstrate that applying deep representation learning is key to unlocking the potential of unsupervised discovery in future datasets from telescopes such as the Vera C. Rubin Observatory and the Square Kilometre Array.

Human review

How Pith checks papers → · Full audit of this paper →

References

How the literature uses it

2 arXiv papers cite this in 2 sentences

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

Papers citing it · 1 sentence without a confident label

Cites it as background 1 paper

More recently, [ref] demonstrated the application of self-supervised representation learning on the Galaxy Zoo DECaLS dataset using Bootstrap Your Own Latent (BYOL) [ref] , which produces low-dimensional embeddings that naturally cluster galaxies based on morphology and facilitate anomaly discovery.

Galaxy Morphology Classification: Uncertainty Modeling and Out of Distribution Detection · Introduction · 0.89

What it cites · 16 sentences without a confident label

Uses the method or tool of 3 papers

We use the package byol-pytorch[<https://github.com/lucidrains/byol-pytorch>] [ref] to apply the BYOL algorithm.

Exploring Simple Siamese Representation Learning · Augmentations and hyperparameters · 0.98
In keeping with the current trend in the machine learning field, we make use of the technique Uniform Manifold Approximation and Projection [ref] .

UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction · Visualisation of extracted features · 0.99
This research made use of the python programming language and the following open source packages: Numpy, SciPy [ref] , Matplotlib [ref] , Seaborn [ref] , scikit-learn [ref] , Pandas [ref] , Astropy [ref] , umap-learn [ref] , PyTorch [ref] and BYOL-PyTorch [ref] .

PyTorch: An Imperative Style, High-Performance Deep Learning Library · Acknowledgements · 1.00

Compares with 1 paper

The overall accuracy is comparable to that in [ref] , if slightly lower because we are using KNN rather than a full CNN for the final classification step.

Radio Galaxy Zoo: Towards building the first multi-purpose foundation model for radio astronomy with self-supervised learning · Evaluation · 1.00

Cites as background 1 paper

Modern approaches to unsupervised clustering of galaxy images tend to rely on autoencoders to learn a representative feature space [ref] .

The Classification of Optical Galaxy Morphology Using Unsupervised Learning Techniques · Introduction · 0.80

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.