Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Self-supervised learning for radio-astronomy source classification: a benchmark

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper shows that self-supervised pretraining on unlabeled radio images beats ImageNet pretraining for radio-source classification.

desk verdict Useful benchmark, but the headline result is partly an artifact of a preprocessing mismatch and the abstract overstates the fine-tuning outcome. read the letter →

arxiv 2411.14078 v2 pith:LEQVHXCH submitted 2024-11-21 astro-ph.IM cs.CV

Thomas Cecconello , Simone Riggi , Ugo Becciani , Fabio Vitello , Andrew M. Hopkins , Giuseppe Vizzari , Concetto Spampinato , Simone Palazzo This is my paper · ORCID
classification astro-ph.IMcs.CV
keywords self-supervisedlearningradioastronomysourceclassificationbenchmarkrepresentationtransferdatacurationinterferometry
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that self-supervised pretraining on radio-interferometry images produces features better suited to radio-source classification than the standard practice of pretraining on optical or natural images. It compares six self-supervised methods (SimCLR, BYOL, DINO, WMSE, SwAV, All4One) pretrained on either a curated set of 17,062 SMGPS source cutouts or an uncurated set of 285,585 sliding-window cutouts, then evaluated on four downstream classification datasets (MiraBest, RGZ, MSRS, VLASS). In linear evaluation, where a frozen backbone feeds a single linear classifier, the radio-pretrained models outperform ImageNet-pretrained BYOL baselines by substantial margins, e.g., 91.0% vs 67.5% on MiraBest. When the whole backbone is fine-tuned, the gap narrows and the advantage is modest. The paper concludes that domain-specific self-supervision, especially with curated data, is a valuable and efficient alternative to natural-image pretraining for the upcoming SKA data deluge.

What carries the argument

The benchmark is built on two pretraining datasets and a fixed evaluation protocol: the Curated dataset (17,062 cutouts centered on SMGPS extended sources, sized at 2.5 times each source bounding box), the Uncurated dataset (285,585 fixed 256x256 sliding-window cutouts from SMGPS and EMU maps), and six augmentation-based SSL objectives (SimCLR, BYOL, DINO, WMSE, SwAV, All4One). Representations are evaluated by 3-fold cross-validation on four downstream classification datasets, in two regimes: linear evaluation (train a linear classifier on frozen backbone features) and fine-tuning (update the whole backbone with AdamW). The load-bearing comparison is linear evaluation, where the frozen-feature accuracy isolates the quality of the pretrained representation.

What would settle it

Run the linear-evaluation protocol again on the same downstream datasets, but perform the balanced resampling only on the training fold after each split (or remove all duplicate cutouts before splitting), and check whether the radio-pretrained models still beat the ImageNet baselines by the same margins; if the margins collapse, the reported advantage is an artifact of train/test leakage.

Watch

Extended reading notes

Core claim

The central claim is that SSL pretraining on radio images yields representations that transfer better to radio-astronomy classification than representations pretrained on natural images, and that this advantage is most visible when the pretrained backbone is used as a fixed feature extractor. The paper supports this with a systematic benchmark: for every SSL method and both backbones, the radio-pretrained models exceed the ImageNet-pretrained baseline on almost all downstream tasks in linear evaluation, with All4One, BYOL, and SimCLR leading. Curated data outperform uncurated data despite being nearly seventeen times smaller, indicating sample quality matters more than volume. The authors interpret the results as evidence that radio-specific pretraining can reduce dependence on large labeled sets and better exploit the abundant unlabeled radio sky.

Load-bearing premise

The reported accuracies assume that the balanced resampling of the downstream datasets is applied inside each cross-validation fold, so duplicated minority-class cutouts never appear in both the training and test partitions; the paper does not state where resampling happens in the pipeline.

Editorial extensions

If this is right

  • Radio-pretrained SSL models can serve as ready-made feature extractors for SKA-era surveys, giving accurate classifications without per-task backbone fine-tuning.
  • Data curation, not just volume, is the main driver of representation quality, so survey teams should invest in catalogue-based cutout extraction.
  • The margins in linear evaluation suggest that future radio-domain foundation models should be pretrained on radio data rather than initialized from ImageNet.
  • MiraBest is nearly saturated in fine-tuning (above 96% for all methods), so future benchmarks should rely on harder tasks like RGZ or MSRS to discriminate between pretraining strategies.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the ranking holds under a stricter evaluation protocol, one immediate extension is to use these SSL backbones for anomaly detection and source-finding, where the same latent representations could flag rare morphologies without labels.
  • The success of curated pretraining suggests a data-engineering recipe for other wavelength regimes: cut out sources from catalogues rather than sampling blindly; the same recipe could apply to optical and infrared surveys.
  • The authors leave multimodal extensions implicit; combining radio cutouts with optical or infrared hosts in a joint SSL objective could produce representations that transfer even better to classification of multi-wavelength source populations.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This paper presents a benchmark of six self-supervised learning methods (SimCLR, BYOL, DINO, WMSE, SwAV, All4One) pretrained on two radio-continuum datasets (a curated SMGPS set and an uncurated SMGPS/EMU set), evaluated by linear probing and fine-tuning on four radio classification tasks (MiraBest, RGZ, MSRS, VLASS), with ImageNet-pretrained BYOL as a baseline. The authors report that radio-pretrained SSL models substantially outperform the ImageNet baseline in linear evaluation, that curated pretraining helps more than dataset size, and that All4One is the strongest linear-evaluation method; they also introduce the MSRS dataset. The headline claim is that SSL pretraining on radio data yields transferable representations for SKA-era analysis.

Significance. The benchmark is timely and useful: it compares six SSL methods in a consistent solo-learn framework, uses public downstream tasks, releases code and models, and introduces MSRS, a multi-survey labeled dataset. The linear-evaluation results are clearly tabulated and the experimental design is largely reproducible. If the main advantage over ImageNet is confirmed under a fair preprocessing protocol, the result would be a valuable practical guide for SKA preparation. At present, however, the headline claim is not fully established because of the normalization mismatch with the ImageNet baseline, an ambiguous resampling protocol, and a confounded curation comparison.

major comments (4)
  1. [Section 4.1] Section 4.1 states that all input images are normalized with per-cutout min-max scaling and then resized to 224x224, but the ImageNet-pretrained BYOL baselines were trained with the standard per-channel mean/std normalization of ImageNet. Under linear evaluation the backbone is frozen, so presenting min-max-normalized inputs to an ImageNet model is a distribution shift that can substantially degrade the extracted features; the model cannot adapt. This is consistent with Table 2, where fine-tuning (which can adapt to the input distribution) removes most of the gap between ImageNet and radio-pretrained models. Consequently the headline linear-evaluation advantage in Table 1, including the flagship MiraBest comparison (SimCLR 91.0 vs ImageNet 67.5), is not a controlled comparison. Please add an ImageNet-pretrained baseline evaluated with ImageNet-style normalization, or otherwise show that the preprocessing mismatch does not change the ranking.
  2. [Section 3.4] Section 3.4 resamples each downstream dataset to balance classes by undersampling or duplicating samples, but it does not state whether this is done before or after the 3-fold cross-validation described in Section 4.1. If resampling precedes the split, duplicated cutouts from the same minority source can appear in both training and test folds, inflating the reported accuracies and potentially changing the SSL-versus-ImageNet ranking. MiraBest is not resampled, so this does not explain that result, but it affects RGZ, MSRS, and VLASS. Please clarify the order, and if resampling was performed before splitting, rerun the evaluation with resampling nested inside each training fold.
  3. [Sections 4.1 and 4.2] Section 4.1 trains the Uncurated models for 100 epochs and the Curated models for 600 epochs, with the same batch size. The curated-vs-uncurated comparison therefore varies curation, dataset size (285,585 vs 17,062), and total optimizer steps simultaneously, so Section 4.2's conclusion that 'dataset curation positively impacts results' and matters more than dataset size is not supported by this design. The claim that 'all SSL methods benefit' is also contradicted by Table 1: DINO on MiraBest drops from 83.6 uncurated to 64.2 curated (and from 69.0 to 66.6 on VLASS). Please match training budgets, or explicitly treat the epoch difference as a confound and weaken the curation claim accordingly.
  4. [Abstract and Section 5] The abstract states that when the entire backbone is fine-tuned, 'the benefits of SSL are less evident but still outperform pretraining,' and Section 5 repeats the claim. Table 2 does not support it: the ImageNet baselines match or exceed the radio-pretrained models in most configurations, e.g., ImageNet-100 BYOL ResNet-18 reaches 96.2 on MiraBest, 81.0 on RGZ, 75.4 on MSRS, and 84.1 on VLASS, compared with best curated ResNet-18 values of 99.2, 81.6, 77.0, and 83.5, respectively; ImageNet-1k BYOL ResNet-50 reaches 98.5 on MiraBest. Please revise the abstract and conclusions to state that fine-tuning benefits are not observed relative to ImageNet pretraining in this benchmark.
minor comments (4)
  1. [Section 3.3] The WMSE description ends with a sentence fragment ('The loss could be represented as: uses the mutual information maximization...'), and the displayed loss only contains the whitening term; please rewrite for clarity.
  2. [Sections 3.3 and 4.1] There are several typos: 'horiziontal' (horizontal), 'temperatur' (temperature), 'redundandy reduction loss' (redundancy reduction loss), and 'abilty' (ability) in Section 3.4.
  3. [Section 4.1] Please state explicitly whether the 3-fold cross-validation folds are stratified by class and how the 9 accuracy values (3 seeds x 3 folds) are combined; the current text reports only that the mean and standard deviation are computed.
  4. [Section 4.2] The text says ImageNet baselines are 'significantly worse,' but no significance tests are reported; please either add appropriate statistical tests or rephrase to 'numerically worse' where the standard deviations overlap.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the benchmark's comparisons are measured against external downstream labels, and the only self-citation (for the pretraining datasets) is a data resource, not a load-bearing premise.

full rationale

The paper is an empirical benchmark, not a derivation. The central claim that SSL-trained models outperform ImageNet-pretrained baselines in linear evaluation is supported by accuracy measurements on external downstream datasets: MiraBest, RGZ, VLASS, and MSRS. Downstream labels come from these public datasets or from a taxonomy newly defined in this paper; they are not produced by the SSL pretraining. SSL hyperparameters are selected on pretraining loss, not on downstream accuracy: Section 4.1 states that 'The selection of other hyperparameters is carried out independently for each SSL method, by manually varying key parameters and observing the average loss on the Curated and Uncurated datasets.' Thus the reported accuracies are not fitted values renamed as predictions. The only self-citation is reference [29], the authors' prior work that constructed the curated and uncurated unlabeled datasets used for pretraining. This is a data resource, not a theorem or an ansatz that forces the results; the evaluations themselves are run independently in this paper. Potential experimental-fairness concerns, such as applying per-cutout min-max normalization to ImageNet-pretrained backbones in linear evaluation, or the partial overlap between MSRS and the uncurated pretraining set, are correctness issues rather than circularity, because the reported numbers are measured rather than derived from the assumptions. No step in the paper reduces by construction to its own inputs.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities or mediators are introduced. The MSRS dataset is a new labeling of existing survey sources, not a new object. The central empirical claim depends on hand-chosen SSL hyperparameters (not fitted to downstream tasks) and on domain assumptions about augmentations, resampling, and label fidelity.

free parameters (5)
  • SSL method learning rates (SimCLR, BYOL, All4One: 1.0-1.2; DINO: 0.016; WMSE: 0.002; SwAV: 1.2) = varies by method, e.g., SimCLR 1.2, DINO 0.016
    Chosen by manually varying and observing pretraining loss (Section 4.1); affects representation quality.
  • Projection sizes per method = 128-512
    Selected per method in Section 4.1.
  • SwAV number of prototypes = 300
    Hyperparameter chosen in Section 4.1.
  • Augmentation transform probabilities = p=0.5 flip, p=0.25 blur, p=0.5 contrast, p=1.0 crop
    Hand-fixed augmentation schedule in Section 4.1.
  • Pretraining epochs per dataset = 100 epochs (Uncurated), 600 epochs (Curated)
    This unequal budget confounds the curation-vs-size comparison (Section 4.1).
assumptions (4)
  • domain assumption Radio images are adequately represented by the chosen set of augmentations (flip, blur, contrast, crop) for SSL.
    SSL success depends on augmentations; the paper does not validate this for radio data beyond empirical results.
  • domain assumption Balanced resampling with duplication of minority samples preserves a fair evaluation.
    Section 3.4 resamples by undersampling or duplicating; if done before CV, duplicates can leak across folds.
  • domain assumption The curated dataset built from SMGPS extended sources is representative of radio source morphology for pretraining.
    Section 3.2 defines the curated dataset; its coverage of the taxonomy is not quantified.
  • domain assumption The new MSRS labels and the existing labels (RGZ, MiraBest, VLASS) are ground-truth correct as provided.
    Downstream accuracies assume label quality; the paper does not audit labels.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Self-supervised learning for radio-astronomy source classification: a benchmark." pith.science (2026). https://pith.science/paper/LEQVHXCH

@misc{pith2026241114078,
  author       = {Pith},
  title        = {Pith review of: Self-supervised learning for radio-astronomy source classification: a benchmark},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LEQVHXCH}},
  note         = {Machine review of arXiv:2411.14078}
}
read the original abstract

The upcoming Square Kilometer Array (SKA) telescope marks a significant step forward in radio astronomy, presenting new opportunities and challenges for data analysis. Traditional visual models pretrained on optical photography images may not perform optimally on radio interferometry images, which have distinct visual characteristics. Self-Supervised Learning (SSL) offers a promising approach to address this issue, leveraging the abundant unlabeled data in radio astronomy to train neural networks that learn useful representations from radio images. This study explores the application of SSL to radio astronomy, comparing the performance of SSL-trained models with that of traditional models pretrained on natural images, evaluating the importance of data curation for SSL, and assessing the potential benefits of self-supervision to different domain-specific radio astronomy datasets. Our results indicate that, SSL-trained models achieve significant improvements over the baseline in several downstream tasks, especially in the linear evaluation setting; when the entire backbone is fine-tuned, the benefits of SSL are less evident but still outperform pretraining. These findings suggest that SSL can play a valuable role in efficiently enhancing the analysis of radio astronomical data. The trained models and code is available at: \url{https://github.com/dr4thmos/solo-learn-radio}

Figures

Figures reproduced from arXiv: 2411.14078 by the authors.

Figure 1
Figure 1. Different visual characteristics of radio images. a) a multi-island radio source in low resolution; [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Images extracted from Curated and Uncurated dataset. Curated samples correspond to well [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Image samples for the downstream datasets employed in our study. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Source Finding and Characterisation for SKAO Science

    astro-ph.IM 2026-07 accept novelty 2.0 of 10

    A review of classical and ML source-finding and morphological classification techniques for SKAO-scale continuum and spectral-line surveys, with emphasis on limitations and pipeline needs.

Reference graph

Works this paper leans on

35 extracted references · 20 canonical work pages · cited by 1 Pith paper

  1. [1]

    arXiv preprint arXiv:2210.07277 (2022)

    Assran, M., Balestriero, R., Duval, Q., Bordes, F., Misra, I., Bojanowski, P., Vin- cent, P., Rabbat, M., Ballas, N.: The hidden uniform cluster prior in self-supervised learning. arXiv preprint arXiv:2210.07277 (2022)

    arXiv 2022

  2. [2]

    Monthly Notices of the Royal Astronomical Society453(3), 2326–2340 (2015)

    Banfield, J.K., Wong, O., Willett, K.W., Norris, R.P., Rudnick, L., Shabala, S.S., Simmons, B.D., Snyder, C., Garon, A., Seymour, N., et al.: Radio galaxy zoo: host galaxies and radio morphologies derived from visual inspection. Monthly Notices of the Royal Astronomical Society453(3), 2326–2340 (2015)

    work page 2015

  3. [3]

    Astrophysical Journal v

    Becker, R.H., White, R.L., Helfand, D.J.: The first survey: faint images of the radio sky at twenty centimeters. Astrophysical Journal v. 450, p. 559450, 559 (1995)

    work page 1995

  4. [4]

    In: ML4Astro International Conference

    Bordiu, C., Bufano, F., Cecconello, T., Sciacca, E., Riggi, S., Vizzari, G.: Patterns in the chaos: An unsupervised view of galactic supernova remnants. In: ML4Astro International Conference. pp. 61–65. Springer (2022)

    work page 2022

  5. [5]

    Advances in neural information processing systems33, 9912–9924 (2020)

    Caron, M., Misra, I., Mairal, J., Goyal, P., Bojanowski, P., Joulin, A.: Unsupervised learning of visual features by contrasting cluster assignments. Advances in neural information processing systems33, 9912–9924 (2020)

    2020

  6. [6]

    In: Proceedings of the International Conference on Computer Vision (ICCV) (2021)

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: Proceedings of the International Conference on Computer Vision (ICCV) (2021)

    2021

  7. [7]

    In: International conference on machine learning

    Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A simple framework for con- trastive learning of visual representations. In: International conference on machine learning. pp. 1597–1607. PMLR (2020)

    2020

  8. [8]

    Evaluating Deep Learning in SystemML using Layer-wise Adaptive Rate Scaling(LARS) Optimizer

    Chowdhury, K., Sharma, A., Chandrasekar, A.D.: Evaluating deep learning in systemml using layer-wise adaptive rate scaling (lars) optimizer. arXiv preprint arXiv:2102.03018 (2021)

    work page Pith review arXiv 2021

Show all 35 references
  1. [9]

    daCosta,V.G.T.,Fini,E.,Nabi,M.,Sebe,N.,Ricci,E.:solo-learn:Alibraryofself- supervisedmethodsforvisualrepresentationlearning.JournalofMachineLearning Research 23(56), 1–6 (2022), http://jmlr.org/papers/v23/21-1155.html

    2022

  2. [10]

    SKA Organisation, Design Report SKA-TEL-SKO-0000002, Rev3 (2016)

    Dewdney, P., et al.: Ska1 system baseline design v2. SKA Organisation, Design Report SKA-TEL-SKO-0000002, Rev3 (2016)

    2016

  3. [11]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

    2020 arXiv

  4. [12]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Dwibedi,D.,Aytar,Y.,Tompson,J.,Sermanet,P.,Zisserman,A.:Withalittlehelp from my friends: Nearest-neighbor contrastive learning of visual representations. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 9588–9597 (2021)

    2021

  5. [13]

    In: International Conference on Machine Learning

    Ermolov, A., Siarohin, A., Sangineto, E., Sebe, N.: Whitening for self-supervised representation learning. In: International Conference on Machine Learning. pp. 3015–3024. PMLR (2021) SSL benchmark for radio-astronomy 15

    2021

  6. [14]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (2023)

    Estepa, I.G., Sarasúa, I., Nagarajan, B., Radeva, P.: All4one: Symbiotic neighbour contrastive learning via self-attention and redundancy reduction. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (2023)

    2023

  7. [15]

    arXiv preprint arXiv:2312.07275 (2023)

    Goedhart, S., Cotton, W., Camilo, F., Thompson, M., Umana, G., Bietenholz, M., Woudt, P., Anderson, L., Bordiu, C., Buckley, D., et al.: The sarao meerkat 1.3 ghz galactic plane survey. arXiv preprint arXiv:2312.07275 (2023)

    2023 arXiv

  8. [16]

    Monthly Notices of the Royal Astronomical Society531(1), 649–681 (2024)

    Goedhart, S., Cotton, W., Camilo, F., Thompson, M., Umana, G., Bietenholz, M., Woudt, P., Anderson, L., Bordiu, C., Buckley, D., et al.: The sarao meerkat 1.3 ghz galactic plane survey. Monthly Notices of the Royal Astronomical Society531(1), 649–681 (2024)

    2024

  9. [17]

    Gordon, Y.A., Rudnick, L., Andernach, H., Morabito, L.K., O’Dea, C.P., Achong, K.M., Baum, S.A., Bayona-Figueroa, C., Hooper, E.J., Mingo, B., et al.: A quick look at the 3 ghz radio sky. ii. hunting for dragns in the vla sky survey. The Astrophysical Journal Supplement Series...

    2023

  10. [18]

    Advances in neural information processing systems 33, 21271–21284 (2020)

    Grill, J.B., Strub, F., Altché, F., Tallec, C., Richemond, P., Buchatskaya, E., Do- ersch, C., Avila Pires, B., Guo, Z., Gheshlaghi Azar, M., et al.: Bootstrap your own latent-a new approach to self-supervised learning. Advances in neural information processing systems 33, 212...

    2020

  11. [19]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.: Masked autoencoders are scalable vision learners. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 16000–16009 (2022)

    2022

  12. [20]

    RAS Techniques and Instruments2(1), 441–452 (2023)

    Huertas-Company, M., Sarmiento, R., Knapen, J.H.: A brief review of contrastive learning applied to astrophysics. RAS Techniques and Instruments2(1), 441–452 (2023)

    2023

  13. [21]

    IEEE Transactions on Knowledge and Data Engineering 35(1), 857–876 (2023)

    Liu, X., Zhang, F., Hou, Z., Mian, L., Wang, Z., Zhang, J., Tang, J.: Self- supervised learning: Generative or contrastive. IEEE Transactions on Knowledge and Data Engineering 35(1), 857–876 (2023). https://doi.org/10.1109/TKDE. 2021.3090866

    2023

  14. [22]

    Astronomy and Computing 36, 100481 (2021)

    Lochner, M., Bassett, B.: Astronomaly: Personalised active anomaly de- tection in astronomical data. Astronomy and Computing 36, 100481 (2021). https://doi.org/https://doi.org/10.1016/j.ascom.2021.100481, https://www.sciencedirect.com/science/article/pii/S2213133721000354

    2021

  15. [23]

    arXiv preprint arXiv:1711.05101 (2017)

    Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101 (2017)

    2017 arXiv

  16. [24]

    Monthly Notices of the Royal Astronomi- cal Society 530(1), 1274–1295 (2024)

    Mohale, K., Lochner, M.: Enabling unsupervised discovery in astronomical images through self-supervised representations. Monthly Notices of the Royal Astronomi- cal Society 530(1), 1274–1295 (2024)

    2024

  17. [25]

    Astronomy & Astrophysics645, A89 (2021)

    Mostert, R.I., Duncan, K.J., Röttgering, H.J., Polsterer, K.L., Best, P.N., Brienza, M., Brüggen, M., Hardcastle, M.J., Jurlin, N., Mingo, B., et al.: Unveiling the rarest morphologies of the lofar two-metre sky survey radio source population with self-organised maps. Astronom...

    2021

  18. [26]

    Publications of the Astronomical Society of Australia28(3), 215–248 (2011)

    Norris, R.P., Hopkins, A.M., Afonso, J., Brown, S., Condon, J.J., Dunne, L., Feain, I., Hollow, R., Jarvis, M., Johnston-Hollitt, M., et al.: Emu: evolutionary map of the universe. Publications of the Astronomical Society of Australia28(3), 215–248 (2011)

    2011

  19. [27]

    arXiv preprint arXiv:2304.07193 (2023)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

    2023 arXiv

  20. [28]

    RAS Techniques and Instruments2(1), 293– 306 (06 2023).https://doi.org/10.1093/rasti/rzad017 16 Cecconello et al

    Porter, F.A.M., Scaife, A.M.M.: MiraBest: a data set of morphologically classified radio galaxies for machine learning. RAS Techniques and Instruments2(1), 293– 306 (06 2023).https://doi.org/10.1093/rasti/rzad017 16 Cecconello et al

    2023 doi

  21. [29]

    arXiv preprint arXiv:2404.18462 (2024)

    Riggi, S., Cecconello, T., Palazzo, S., Hopkins, A., Gupta, N., Bordiu, C., In- gallinera, A., Buemi, C., Bufano, F., Cavallaro, F., et al.: Self-supervised con- trastive learning of radio data for source detection, classification and peculiar ob- ject discovery. arXiv preprin...

    2024 arXiv

  22. [30]

    Astronomy and Computing 42, 100682 (2023)

    Riggi, S., Magro, D., Sortino, R., De Marco, A., Bordiu, C., Cecconello, T., Hop- kins,A.M.,Marvil,J.,Umana,G.,Sciacca,E.,etal.:Astronomicalsourcedetection in radio continuum maps with deep neural networks. Astronomy and Computing 42, 100682 (2023)

    2023

  23. [31]

    Publications of the Astronomical Society of Australia41, e029 (2024)

    Riggi,S.,Umana,G.,Trigilio,C.,Bordiu,C.,Bufano,F.,Ingallinera,A.,Cavallaro, F., Gordon, Y., Norris, R.P., Gürkan, G., et al.: Classification of compact radio sources in the galactic plane with supervised machine learning. Publications of the Astronomical Society of Australia41...

    2024

  24. [32]

    The Astrophysical Journal921(2), 177 (2021)

    Sarmiento, R., Knapen, J.H., Sánchez, S.F., Sánchez, H.D., Drory, N., Falcón- Barroso, J., et al.: Capturing the physics of manga galaxies with self-supervised machine learning. The Astrophysical Journal921(2), 177 (2021)

    2021

  25. [33]

    RAS Techniques and Instruments 3(1), 19–32 (12 2023)

    Slijepcevic, I.V., Scaife, A.M.M., Walmsley, M., Bowles, M., Wong, O.I., Shabala, S.S., White, S.V.: Radio galaxy zoo: towards building the first multipurpose foun- dation model for radio astronomy with self-supervised learning. RAS Techniques and Instruments 3(1), 19–32 (12 2023)

    2023

  26. [34]

    Monthly Notices of the Royal Astronomical Society 454(1), 902–912 (2015)

    Umana, G., Trigilio, C., Franzen, T., Norris, R.P., Leto, P., Ingallinera, A., Buemi, C.S., Agliozzo, C., Cavallaro, F., Cerrigone, L.: Scorpio: a deep survey of radio emission from the stellar life-cycle. Monthly Notices of the Royal Astronomical Society 454(1), 902–912 (2015)

    2015

  27. [35]

    In: International Conference on Machine Learn- ing

    Zbontar, J., Jing, L., Misra, I., LeCun, Y., Deny, S.: Barlow twins: Self-supervised learning via redundancy reduction. In: International Conference on Machine Learn- ing. pp. 12310–12320. PMLR (2021)

    2021

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.