Pith. sign in

REVIEW 6 minor 161 references

SKAO-scale radio surveys need automated hybrid finders because classical tools alone cannot handle the volume, dynamic range and morphological diversity of sources.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-12 00:17 UTC pith:R3Q5G3AH

load-bearing objection Solid AASKA-II review chapter that organises the real operational bottlenecks for SKAO catalogues; no new algorithm or result, but accurate and useful for the community.

arxiv 2607.03736 v1 pith:R3Q5G3AH submitted 2026-07-04 astro-ph.IM astro-ph.GA

Source Finding and Characterisation for SKAO Science

Sabyasachi Pal , Souvik Manik , M Carmen Toribio , Geferson Lucatelli , Omkar Bait , Simone Riggio , Antxon Alberdi , Philippa Hartley
show 3 more authors
Javier Moldon Mamta Pandey-Pommier Rob Beswick
This is my paper
classification astro-ph.IM astro-ph.GA
keywords source findingSKAOradio continuummachine learningmorphological classificationspectral-line surveysHI emissionradio galaxies
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This review chapter argues that the Square Kilometre Array Observatory and its precursors will deliver the deepest radio images ever made, revealing billions of sources whose complex shapes and faint diffuse emission make manual cataloguing impossible. Classical source finders that estimate background noise, detect islands of emission and fit Gaussians remain useful for compact objects but systematically miss or fragment extended, low-surface-brightness and multi-component structures. Machine-learning methods, especially convolutional networks and U-Nets, recover more of those difficult sources and can operate on both continuum images and spectral-line cubes, yet they still depend on realistic training data and struggle with artefacts, RFI and instrument-to-instrument differences. The authors therefore conclude that continued algorithmic advances, hybrid classical-plus-ML pipelines and matching computational infrastructure are essential if SKAO is to convert its raw data into reliable science-ready catalogues. Without them, incompleteness and bias will propagate into luminosity functions, clustering statistics and studies of galaxy evolution.

Core claim

The paper establishes that classical threshold-and-fit source finders, while still the workhorses of radio survey pipelines, cannot simultaneously deliver the completeness, reliability and morphological fidelity required by SKAO-scale data; machine-learning architectures and hybrid end-to-end pipelines are therefore indispensable for detecting and characterising compact, extended, diffuse and transient radio sources amid realistic noise and artefacts.

What carries the argument

The three-stage classical pipeline (local background/noise estimation, island detection, multi-component Gaussian or Sérsic modelling) combined with deep-learning segmenters and classifiers (CNNs, U-Nets, Mask R-CNN, YOLO-style detectors) that learn non-linear patterns directly from images or even visibilities; hybrid integration of both is presented as the practical path forward.

Load-bearing premise

That sufficiently realistic labelled training sets, or self-supervised proxies, can be generated so that deep-learning finders and classifiers will generalise across the heterogeneous noise, resolution and artefact properties of SKAO and its precursors.

What would settle it

If large SKA Data Challenges or early SKAO continuum and HI surveys demonstrate that carefully tuned classical finders alone reach completeness and reliability within a few percent of hybrid ML methods for both compact and diffuse sources across the full dynamic range, the claim that continued algorithmic and infrastructure advances are essential would be directly weakened.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • SKAO catalogues will be complete enough for unbiased source counts, luminosity functions and clustering statistics.
  • Improved recovery of faint extended emission will open systematic studies of radio relics, halos and bent-tailed sources as probes of cluster physics and cosmic magnetism.
  • Hybrid pipelines will enable real-time or near-real-time classification of transients, supporting multi-messenger follow-up.
  • End-to-end automated workflows will become the default path from raw visibilities to science-ready catalogues, reducing human intervention to rare or ambiguous cases.
  • Rare morphological classes such as giant radio galaxies and odd radio circles will be inventoried at scale rather than by chance discovery.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The performance gap between classical and ML finders will shrink only if the community builds shared, interferometrically realistic simulation suites that include direction-dependent calibration residuals and polarisation.
  • Polarisation and multi-frequency spectral cubes will need to become first-class inputs to source characterisation rather than optional post-processing layers.
  • If computational investment lags behind imaging capability, source finding itself—not raw data volume—could become the rate-limiting step for SKAO science delivery.
  • Citizen-science and expert labels will require soft or probabilistic treatment to avoid injecting annotator noise into the next generation of supervised classifiers.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 6 minor

Summary. This AASKA-II review chapter surveys classical and machine-learning methods for radio continuum and spectral-line source finding and morphological characterisation in the SKAO era. It summarises the three-stage classical pipeline (background/noise estimation, island detection, component modelling), 3D HI tools (SoFiA, DUCHAMP, FLASHfinder), parametric multi-component Sérsic decomposition, non-parametric CASGM metrics, and a range of CNN/U-Net/object-detection and weakly supervised classifiers. Drawing on SKA Data Challenges and precursor surveys (LoTSS, EMU, MIGHTEE, WALLABY, etc.), it argues that classical finders alone cannot meet the volume, dynamic range and morphological diversity expected from SKAO, so hybrid ML pipelines, scalable infrastructure and continued algorithmic development are essential.

Significance. As a synthesis chapter the paper does not claim new algorithms or quantitative results, but it provides a timely, well-referenced map of the tools, failure modes (diffuse emission, artefacts, domain shift, training-set bias) and open challenges that will shape SKAO catalogue production. The explicit treatment of HI emission/absorption, multi-scale/multi-wavelength decomposition, and the limits of supervised ML generalisation is useful for both instrument teams and science users. Strengths include the clear classical-pipeline taxonomy, Table 1, the balanced discussion of training-data and domain-adaptation issues in §§2.3–4, and the forward-looking hybrid-pipeline recommendations. The central claim is literature-supported and proportionate for a review.

minor comments (6)
  1. Table 1 is incomplete relative to the surrounding text: multiprocessing ticks are missing for several packages that the prose states support parallelisation (e.g. Aegean, Selavy), and the compact/extended/diffuse columns for some tools (e.g. PyBDSF, ProFound) do not fully match the descriptive claims in §2.1. A short consistency pass would help.
  2. Figure 1 is referenced as illustrating the PyBDSF stages but is not described in the caption or body beyond a one-line mention; a brief caption that names the three stages would improve standalone readability.
  3. §3.3.1 Eq. (1)–(2): the Sérsic multi-component model is standard, but a one-sentence note on how beam convolution is (or is not) included in the fit would avoid ambiguity for radio readers.
  4. Scattered typographical issues (missing spaces after commas/periods in the abstract and early sections, duplicated Bertin & Arnouts 1996 reference, occasional hyphenation artefacts such as “thedevelopmentofautomatedalgorithms”) should be cleaned in production.
  5. §2.3 and §3.4 cite many recent ML detectors (ContinUNet, YOLO-CIANNA, RadioGalaxyNET, etc.); a compact summary table of architecture, training regime and reported completeness/reliability on a common benchmark (e.g. SDC) would make the comparative claims easier to navigate.
  6. The abstract lists MWA and HERA among precursors that “will enable us to create the deepest radio images,” yet the body focuses almost exclusively on continuum/HI continuum-pathfinder experience; a short clause clarifying the intended scope (or a pointer to 21-cm intensity-mapping source-finding) would avoid over-promising.

Circularity Check

0 steps flagged

No significant circularity: literature review with no novel derivation, fitted prediction, or load-bearing self-referential claim.

full rationale

This AASKA-II chapter is a survey of classical source finders (Aegean, PyBDSF, SoFiA, Caesar, etc.), 3D spectral-line tools, ML architectures (U-Net, Mask R-CNN, YOLO variants), morphological classification schemes (FR I/II, HyMoRS, GRGs, ORCs, CASGM, multi-component Sérsic), limitations (artefacts, domain shift, training-data scarcity), and future hybrid pipelines. It advances no new quantitative result, theorem, uniqueness claim, or fitted parameter that is then re-presented as a prediction. Self-citations (e.g., Riggi et al. on Caesar, Lucatelli et al. on multi-component fitting, Pal/Kumari/Manik/Bhukta works on morphologies) supply illustrative examples and prior tool descriptions; they are not the sole support for any central premise that would collapse without them. The core assertion—that classical finders alone cannot meet SKAO-scale volume, dynamic range and morphological diversity, so algorithmic and infrastructure advances remain essential—is grounded in the external literature (Hopkins et al. 2015, Hancock et al., Bonaldi et al. SKA Data Challenges, Serra et al. SoFiA, etc.) and is therefore self-contained. No equation reduces to its own inputs by construction, no ansatz is smuggled via self-citation, and no known empirical pattern is merely renamed. Score 0 is therefore the correct, proportionate finding.

Axiom & Free-Parameter Ledger

0 free parameters · 3 axioms · 0 invented entities

As a review the paper inherits the standard assumptions of radio interferometry and the published performance claims of the tools it surveys; it introduces no free parameters of its own and no new physical entities.

axioms (3)
  • domain assumption Radio interferometric images contain correlated noise, residual sidelobes and direction-dependent artefacts that can mimic or mask real sources.
    Stated throughout §§1–2 and 4 as the reason classical thresholding fails; taken as established from the interferometry literature.
  • domain assumption Completeness, reliability and parameter accuracy are the three primary metrics by which source finders must be judged.
    Introduced in §1 and used as the evaluation framework for all subsequent comparisons.
  • domain assumption Supervised ML performance is limited by the realism and diversity of labelled training sets relative to real SKAO data.
    Repeated in §§2.3, 3.4 and 4; treated as a community consensus rather than a new claim.

pith-pipeline@v1.1.0-grok45 · 32098 in / 1918 out tokens · 18376 ms · 2026-07-12T00:17:07.845552+00:00 · methodology

0 comments
read the original abstract

The advancements in highly sensitive and powerful radio telescopes, including the Square Kilometre Array Observatory (SKAO) and its precursors, MeerKAT, ASKAP, MWA, and HERA, will enable us to create the deepest radio images of the sky. However, due to the sheer scale of the datasets, manually classifying and analyzing this data is computationally expensive, time-consuming, and laborious. Therefore, the development of automated algorithms to detect and classify complex morphological radio sources from large astronomical surveys is the need of the time. In this chapter, we examine the recent advancements and challenges in source-finding techniques triggered by the analysis of SKAO precursor data and the SKA Data Challenges, both in spectral and continuum modes, along with the growing demand for computational resources and automated source detection methods using machine learning (ML) algorithms for the identification and characterisation of new populations of sources, addressing their complex and diffuse morphologies, as well as transient nature. Additionally, we discuss the critical factors affecting the quality and limitations of automated source-finding techniques, including artefacts from residual continuum emission, sidelobes, radio frequency interference (RFI), technical failures, calibration issues, flagging methods and false positives. With the availability of the full operational phase of the SKAO, continued advancements in source detection algorithms and computational infrastructure will be essential to fully exploit its scientific potential.

Figures

Figures reproduced from arXiv: 2607.03736 by Antxon Alberdi, Geferson Lucatelli, Javier Moldon, Mamta Pandey-Pommier, M Carmen Toribio, Omkar Bait, Philippa Hartley, Rob Beswick, Sabyasachi Pal, Simone Riggio, Souvik Manik.

Figure 1
Figure 1. Figure 1: Different stages of the process of characterisation of radio sources with widely used PyBDSF. 2.1.1 Background and noise estimation A reliable background model is crucial, as it directly affects the sensitivity and false detection rate of the pipeline. Many algorithms, such as Aegean, PyBDSF, and Selavy, use a sliding-box approach in which the local background and noise levels are estimated using neighbori… view at source ↗
Figure 2
Figure 2. Figure 2: Basic building blocks of a radio galaxy classifier based on a convolutional neural network (CNN). 3.4.1.2 Regularization Techniques One of the major challenges in deep learning for radio astronomy is overfitting, primarily due to limited labeled data. To address this, various regularization methods are employed to improve model robustness and prevent memorization of the training set. A widely used approach… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

161 extracted references · 51 canonical work pages · 43 internal anchors

  1. [1]

    Challenging interferometric imaging: Machine learning-based source localization from uv-plane observations

    Challenging interferometric imaging: Machine learning-based source localization from uv-plane observations. , keywords =. doi:https://doi.org/10.1051/0004-6361/202245778 , archivePrefix =. 2305.03533 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202245778

  2. [2]

    Radio-astronomical Image Reconstruction with Conditional Denoising Diffusion Model

    Radio-astronomical image reconstruction with a conditional denoising diffusion model. , keywords =. doi:https://doi.org/10.1051/0004-6361/202347948 , archivePrefix =. 2402.10204 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202347948

  3. [3]

    The ASKAP/EMU Source Finding Data Challenge

    The ASKAP/EMU Source Finding Data Challenge. , keywords =. doi:https://doi.org/10.1017/pasa.2015.37 , archivePrefix =. 1509.03931 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1017/pasa.2015.37 2015

  4. [4]

    Deep Learning Based Detection of Cosmological Diffuse Radio Sources

    Deep learning based detection of cosmological diffuse radio sources. , keywords =. doi:https://doi.org/10.1093/mnras/sty2102 , archivePrefix =. 1809.03315 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/sty2102

  5. [5]

    , keywords =

    DEEPSOURCE: point source detection using deep learning. , keywords =. doi:https://doi.org/10.1093/mnras/stz131 , archivePrefix =. 1807.02701 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stz131

  6. [6]

    ConvoSource: Radio-Astronomical Source-Finding with Convolutional Neural Networks

    ConvoSource: Radio-Astronomical Source-Finding with Convolutional Neural Networks. Galaxies , keywords =. doi:https://doi.org/10.3390/galaxies8010003 , archivePrefix =. 1910.03631 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.3390/galaxies8010003 1910

  7. [7]

    Proceedings of the IEEE international conference on computer vision , pages=

    Mask r-cnn , author=. Proceedings of the IEEE international conference on computer vision , pages=

  8. [8]

    International Conference on Medical image computing and computer-assisted intervention , pages=

    U-net: Convolutional networks for biomedical image segmentation , author=. International Conference on Medical image computing and computer-assisted intervention , pages=. 2015 , organization=

    2015

  9. [9]

    International Workshop on Artificial Intelligence and Pattern Recognition , pages=

    Semantic segmentation of radio-astronomical images , author=. International Workshop on Artificial Intelligence and Pattern Recognition , pages=. 2021 , organization=

    2021

  10. [10]

    , keywords =

    Compact continuum source finding for next generation radio surveys. , keywords =. doi:https://doi.org/10.1111/j.1365-2966.2012.20768.x , archivePrefix =. 1202.4500 , primaryClass =

    doi:10.1111/j.1365-2966.2012.20768.x 2012

  11. [11]

    , keywords =

    Source Finding in the Era of the SKA (Precursors): Aegean 2.0. , keywords =. doi:https://doi.org/10.1017/pasa.2018.3 , archivePrefix =. 1801.05548 , primaryClass =

    Pith/arXiv arXiv doi:10.1017/pasa.2018.3 2018

  12. [12]

    PyBDSF: Python Blob Detection and Source Finder

  13. [13]

    , keywords =

    SOFIA: a flexible source finder for 3D spectral line data. , keywords =. doi:https://doi.org/10.1093/mnras/stv079 , archivePrefix =. 1501.03906 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stv079

  14. [14]

    , keywords =

    SOFIA 2 - an automated, parallel H I source finding pipeline for the WALLABY survey. , keywords =. doi:https://doi.org/10.1093/mnras/stab1881 , archivePrefix =. 2106.15789 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stab1881

  15. [15]

    , keywords =

    DUCHAMP: a 3D source finder for spectral-line data. , keywords =. doi:10.1111/j.1365-2966.2012.20548.x , archivePrefix =. 1201.2710 , primaryClass =

    doi:10.1111/j.1365-2966.2012.20548.x 2012

  16. [16]

    Publications of the Astronomical Society of the Pacific , volume=

    Compact continuum source finding for next generation radio surveys , author=. Publications of the Astronomical Society of the Pacific , volume=. 2005 , doi=

    2005

  17. [17]

    You Only Look Once: Unified, Real-Time Object Detection , year=

    Redmon, Joseph and Divvala, Santosh and Girshick, Ross and Farhadi, Ali , booktitle=. You Only Look Once: Unified, Real-Time Object Detection , year=

  18. [18]

    Deblending and Classifying Astronomical Sources with Mask R-CNN Deep Learning

    Deblending and classifying astronomical sources with Mask R-CNN deep learning. , keywords =. doi:https://doi.org/10.1093/mnras/stz2845 , archivePrefix =. 1908.02748 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stz2845 1908

  19. [19]

    Astronomy and Computing , keywords =

    Radio sources segmentation and classification with deep learning. Astronomy and Computing , keywords =. doi:https://doi.org/10.1016/j.ascom.2023.100728 , archivePrefix =. 2306.01426 , primaryClass =

    doi:10.1016/j.ascom.2023.100728 2023

  20. [20]

    Astronomy and Computing , keywords =

    Astronomical source detection in radio continuum maps with deep neural networks. Astronomy and Computing , keywords =. doi:https://doi.org/10.1016/j.ascom.2022.100682 , archivePrefix =. 2212.02538 , primaryClass =

    doi:10.1016/j.ascom.2022.100682 2022

  21. [21]

    , keywords =

    RadioGalaxyNET: Dataset and novel computer vision algorithms for the detection of extended radio galaxies and infrared hosts. , keywords =. doi:https://doi.org/10.1017/pasa.2023.64 , archivePrefix =. 2312.00306 , primaryClass =

    Pith/arXiv arXiv doi:10.1017/pasa.2023.64 2023

  22. [22]

    YOLO-CIANNA: Galaxy detection with deep learning in radio data. I. A new YOLO-inspired source detection method applied to the SKAO SDC1

    YOLO-CIANNA: Galaxy detection with deep learning in radio data: I. A new YOLO-inspired source detection method applied to the SKAO SDC1. , keywords =. doi:https://doi.org/10.1051/0004-6361/202449548 , archivePrefix =. 2402.05925 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202449548

  23. [23]

    Monthly Notices of the Royal Astronomical Society , volume=

    Rapid sorting of radio galaxy morphology using Haralick features , author=. Monthly Notices of the Royal Astronomical Society , volume=. doi:https://doi.org/10.1093/mnras/stab271 , year=

    work page doi:10.1093/mnras/stab271

  24. [24]

    The Astronomical Journal , volume=

    Morphological-based classifications of radio galaxies using supervised machine-learning methods associated with image moments , author=. The Astronomical Journal , volume=. 2021 , doi =

    2021

  25. [25]

    , keywords =

    Automated detection of extended sources in radio maps: progress from the SCORPIO survey. , keywords =

  26. [26]

    CAESAR source finder: recent developments and testing

    Cuc(aesar) source finder: Recent developments and testing. , keywords =. doi:https://doi.org/10.1017/pasa.2019.29 , archivePrefix =. 1909.06116 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1017/pasa.2019.29 2019

  27. [27]

    , keywords =

    ProFound: Source Extraction and Application to Modern Survey Data. , keywords =. doi:https://doi.org/10.1093/mnras/sty440 , archivePrefix =. 1802.00937 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/sty440

  28. [28]

    Radio Source Extraction with ProFound

    Radio source extraction with PROFOUND. , keywords =. doi:https://doi.org/10.1093/mnras/stz1462 , archivePrefix =. 1902.01440 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stz1462 1902

  29. [29]

    Astronomy and Computing , keywords =

    PySE: Software for extracting sources from radio images. Astronomy and Computing , keywords =. doi:https://doi.org/10.1016/j.ascom.2018.02.003 , archivePrefix =. 1802.09604 , primaryClass =

    Pith/arXiv arXiv doi:10.1016/j.ascom.2018.02.003 2018

  30. [30]

    , keywords =

    BLOBCAT: software to catalogue flood-filled blobs in radio images of total intensity and linear polarization. , keywords =. doi:https://doi.org/10.1111/j.1365-2966.2012.21373.x , archivePrefix =. 1205.5313 , primaryClass =

    doi:10.1111/j.1365-2966.2012.21373.x 2012

  31. [31]

    and Vielva, P

    López-Caniego, M. and Vielva, P. , title=. , year=

  32. [32]

    and Herranz, D

    López-Caniego, M. and Herranz, D. and González-Nuevo, J. and Sanz, J. L. and Barreiro, R. B. and Vielva, P. and Argüeso, F. and Toffolatti, L. , title=. , year=

  33. [33]

    , keywords =

    A Catalog of 1.4 GHz Radio Sources from the FIRST Survey. , keywords =. doi:https://doi.org/10.1086/303564 , adsurl =

    doi:10.1086/303564

  34. [34]

    , volume=

    Source-Finding for the Australian Square Kilometre Array Pathfinder , author=. , volume=. 2012 , doi =

    2012

  35. [35]

    and Arnouts, S

    Bertin, E. and Arnouts, S. , title=. Astronomy & Astrophysics Suppllement Series , year=

  36. [36]

    Dissertation, Astronomical Institute Anton Pannekoek, University of Amsterdam , volume=

    Search and detection of low frequency radio transients , author=. Dissertation, Astronomical Institute Anton Pannekoek, University of Amsterdam , volume=

  37. [37]

    and Staley, Tim D

    Swinbank, John D. and Staley, Tim D. and Molenaar, Gijs J. and Rol, Evert and Rowlinson, Antonia and Scheers, Bart and Spreeuw, Hanno and Bell, Martin E. and Broderick, Jess W. and Carbone, Dario and Garsden, Hugh and van der Horst, Alexander J. and Law, Casey J. and Wise, Michael and Breton, Rene P. and Cendes, Yvette and Corbel, Stéphane and Eislöffel, ...

    2015

  38. [39]

    WALLABY -- An SKA Pathfinder HI Survey

    WALLABY an SKA Pathfinder H I survey. , keywords =. doi:https://doi.org/10.1007/s10509-020-03831-4 , archivePrefix =. 2002.07311 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1007/s10509-020-03831-4 2002

  39. [40]

    Observing, mapping and source extraction , author=

    10C survey of radio sources at 15.7 GHz - I. Observing, mapping and source extraction , author=. , volume=. 2011 , doi =

    2011

  40. [41]

    , pages=

    Square Kilometre Array Science Data Challenge 1: analysis and results , author=. , pages=. 2021 , doi =

    2021

  41. [42]

    Proceedings of Science , year=

    LOFAR Transients and the Radio Sky Monitor , author=. Proceedings of Science , year=

  42. [43]

    , volume=

    LOFAR: The LOw-Frequency ARray , author=. , volume=. 2013 , pages=

    2013

  43. [44]

    , keywords =

    Radio Galaxy Zoo: host galaxies and radio morphologies derived from visual inspection. , keywords =. doi:https://doi.org/10.1093/mnras/stv1688 , archivePrefix =. 1507.07272 , primaryClass =

    Pith/arXiv arXiv doi:10.1093/mnras/stv1688

  44. [45]

    , keywords =

    Feature guided training and rotational standardization for the morphological classification of radio galaxies. , keywords =

  45. [46]

    Astrophysics Source Code Library , pages=

    Pink: Parallelized rotation and flipping invariant kohonen maps , author=. Astrophysics Source Code Library , pages=

  46. [47]

    , keywords =

    Radio Galaxy Zoo: compact and extended radio source classification with deep learning. , keywords =

  47. [48]

    , keywords =

    The FIRST Classifier: compact and extended radio galaxy classification using deep Convolutional Neural Networks. , keywords =

  48. [49]

    , keywords =

    Classifying Radio Galaxies with the Convolutional Neural Network. , keywords =

  49. [50]

    , keywords =

    CNN architecture comparison for radio galaxy classification. , keywords =

  50. [51]

    , keywords =

    Radio Galaxy Zoo: giant radio galaxy classification using multidomain deep learning. , keywords =

  51. [52]

    , keywords =

    Fanaroff-Riley classification of radio galaxies using group-equivariant convolutional neural networks. , keywords =

  52. [53]

    International conference on machine learning , pages=

    Group equivariant convolutional networks , author=. International conference on machine learning , pages=. 2016 , organization=

    2016

  53. [54]

    Advances in neural information processing systems , volume=

    Imagenet classification with deep convolutional neural networks , author=. Advances in neural information processing systems , volume=

  54. [55]

    , keywords =

    Attention-gating for improved radio galaxy classification. , keywords =

  55. [56]

    RAS Techniques and Instruments , keywords =

    Morphological classification of radio galaxies with Wasserstein generative adversarial network-supported augmentation. RAS Techniques and Instruments , keywords =

  56. [57]

    Radio Galaxy Zoo: Machine learning for radio source host galaxy cross-identification

    Radio Galaxy Zoo: machine learning for radio source host galaxy cross-identification. , keywords =. doi:https://doi.org/10.1093/mnras/sty1308 , archivePrefix =. 1805.05540 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/sty1308

  57. [58]

    Artificial intelligence for celestial object census: the latest technology meets the oldest science

    Artificial intelligence for celestial object census: the latest technology meets the oldest science. Science Bulletin , keywords =. doi:https://doi.org/10.1016/j.scib.2021.07.015 , archivePrefix =. 2107.03082 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1016/j.scib.2021.07.015 2021

  58. [59]

    and Riggi, S

    Magro, D. and Riggi, S. and. ASGARD: A deep neural network for the detection of compact sources, extended galaxies, and sidelobes in radio astronomical maps -- In preparation , YEAR =

  59. [60]

    and Riggi, S

    Magro, D. and Riggi, S. and. ASGARD: A deep neural network for the detection of compact sources, extended galaxies, and sidelobes in radio astronomical maps , NOTE =

  60. [61]

    , volume=

    The Completeness and Reliability of Threshold and False-discovery Rate Source Extraction Algorithms for Compact Continuum Sources , author=. , volume=. 2012 , doi=

    2012

  61. [62]

    , volume=

    Photometry of a complete sample of faint galaxies , author=. , volume=. 1980 , pages=

    1980

  62. [63]

    Monthly Notices of the Royal Astronomical Society , volume=

    Data-efficient classification of radio galaxies , author=. Monthly Notices of the Royal Astronomical Society , volume=. 2022 , publisher=

    2022

  63. [64]

    , keywords =

    Transfer learning for radio galaxy classification. , keywords =

  64. [65]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Inception-v4, inception-resnet and the impact of residual connections on learning , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  65. [66]

    2023 International Joint Conference on Neural Networks (IJCNN) , pages=

    Morphological classification of extragalactic radio sources using gradient boosting methods , author=. 2023 International Joint Conference on Neural Networks (IJCNN) , pages=. 2023 , organization=

    2023

  66. [67]

    , keywords =

    Morphological classification of radio galaxies: capsule networks versus convolutional neural networks. , keywords =

  67. [68]

    Astrophys

    Surface Brightness and Evolution of Galaxies , author=. Astrophys. J. Lett. , year=

  68. [69]

    A comparative study of source-finding techniques in HI emission line cubes using SoFiA, MTObjects, and supervised deep learning

    A comparative study of source-finding techniques in H I emission line cubes using SoFiA, MTObjects, and supervised deep learning. , keywords =. doi:https://doi.org/10.1051/0004-6361/202244708 , archivePrefix =. 2211.12809 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202244708

  69. [70]

    SKA Science Data Challenge 2: analysis and results

    SKA Science Data Challenge 2: analysis and results. Monthly Notices of the Royal Astronomical Society , keywords =. doi:https://doi.org/10.1093/mnras/stad1375 , archivePrefix =. 2303.07943 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stad1375

  70. [71]

    Astronomy & Astrophysics , volume=

    Utilization of convolutional neural networks for H I source finding-Team FORSKA-Sweden approach to SKA Data Challenge 2 , author=. Astronomy & Astrophysics , volume=. 2023 , doi =

    2023

  71. [72]

    , keywords =

    Extragalactic radio sources with hybrid morphology: implications for the Fanaroff-Riley dichotomy. , keywords =. doi:https://doi.org/10.48550/arXiv.astro-ph/0009441 , archivePrefix =. astro-ph/0009441 , primaryClass =

    Pith/arXiv arXiv doi:10.48550/arxiv.astro-ph/0009441

  72. [73]

    Unveiling the cause of hybrid morphology radio sources (HyMoRS)

    Unveiling the cause of hybrid morphology radio sources (HyMoRS). , keywords =. doi:https://doi.org/10.1093/mnras/stz3069 , archivePrefix =. 1910.12857 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stz3069 1910

  73. [74]

    Search for hybrid morphology radio galaxies from FIRST survey at 1400 MHz

    Search for hybrid morphology radio galaxies from the FIRST survey at 1400 MHz. , keywords =. doi:https://doi.org/10.1093/mnras/stac1215 , archivePrefix =. 2104.14469 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stac1215

  74. [75]

    , year =

    Manik, Souvik and Kumari, Shobha and Bhukta, Netai and Pal, Sabyasachi and Mondal, Sushanta K. , year =. Hybrid Morphology Radio Sources from the MeerKAT Absorption Line Survey (MALS): Radio, Mid-infrared, and Environmental Characteristics , volume =. The Astrophysical Journal , publisher =. doi:https://doi.org/10.3847/1538-4357/ae1f85 , number =

    doi:10.3847/1538-4357/ae1f85

  75. [76]

    , keywords =

    Source extraction and photometry for the far-infrared and sub-millimeter continuum in the presence of complex backgrounds. , keywords =. doi:https://doi.org/10.1051/0004-6361/201014752 , archivePrefix =. 1011.3946 , primaryClass =

    Pith/arXiv arXiv doi:10.1051/0004-6361/201014752

  76. [77]

    Self-supervised contrastive learning of radio data for source detection, classification and peculiar object discovery

    Self-supervised contrastive learning of radio data for source detection, classification and peculiar object discovery. , keywords =. doi:https://doi.org/10.1017/pasa.2024.84 , archivePrefix =. 2404.18462 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1017/pasa.2024.84 2024

  77. [78]

    Self-Supervised Learning on MeerKAT Wide-Field Continuum Images

    Self-supervised learning on MeerKAT wide-field continuum images. , keywords =. doi:https://doi.org/10.1051/0004-6361/202449964 , archivePrefix =. 2408.06147 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1051/0004-6361/202449964

  78. [79]

    Deep Learning for Morphological Identification of Extended Radio Galaxies using Weak Labels

    Deep learning for morphological identification of extended radio galaxies using weak labels. , keywords =. doi:https://doi.org/10.1017/pasa.2023.46 , archivePrefix =. 2308.05166 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1017/pasa.2023.46 2023

  79. [80]

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

    Radio Galaxy Zoo: using semi-supervised learning to leverage large unlabelled data sets for radio galaxy classification under data set shift. , keywords =. doi:https://doi.org/10.1093/mnras/stac1135 , archivePrefix =. 2204.08816 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/mnras/stac1135

  80. [81]

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

    Radio galaxy zoo: towards building the first multipurpose foundation model for radio astronomy with self-supervised learning. RAS Techniques and Instruments , keywords =. doi:https://doi.org/10.1093/rasti/rzad055 , archivePrefix =. 2305.16127 , primaryClass =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.1093/rasti/rzad055

Showing first 80 references.