CoChat
Sign Up
Cited by 113Galaxies: Formation, Evolution, PhenomenaAdvanced Neural Network ApplicationsAstrophysical Phenomena and Observations

Deep learning predictions of galaxy merger stage and the importance of observational realism

Open access
Bottrell, Connor, Hani, Maan H, Teimoorinia, Hossen, Ellison, Sara L, Moreno, Jorge, Torrey, Paul, Hayward, Christopher C, Thorp, Mallory +2 more·2019·Monthly Notices of the Royal Astronomical Society
DOI 10.1093/mnras/stz2934
Open this paper in CoChat — free accountOpen PDF

Abstract

Machine learning is becoming a popular tool to quantify galaxy morphologies and identify mergers. However, this technique relies on using an appropriate set of training data to be successful. By combining hydrodynamical simulations, synthetic observations, and convolutional neural networks (CNNs), we quantitatively assess how realistic simulated galaxy images must be in order to reliably classify mergers. Specifically, we compare the performance of CNNs trained with two types of galaxy images, stellar maps and dust-inclusive radiatively transferred images, each with three levels of observational realism: (1) no observational effects (idealized images), (2) realistic sky and point spread function (semirealistic images), and (3) insertion into a real sky image (fully realistic images). We find that networks trained on either idealized or semireal images have poor performance when applied to survey-realistic images. In contrast, networks trained on fully realistic images achieve 87.1 per cent classification performance. Importantly, the level of realism in the training images is much more important than whether the images included radiative transfer, or simply used the stellar maps ($87.1{{\ \rm per\ cent}}$ compared to $79.6{{\ \rm per\ cent}}$ accuracy, respectively). Therefore, one can avoid the large computational and storage cost of running radiative transfer with a relatively modest compromise in classification performance. Making photometry-based networks insensitive to colour incurs a very mild penalty to performance with survey-realistic data ($86.0{{\ \rm per\ cent}}$ with r-only compared to $87.1{{\ \rm per\ cent}}$ with gri). This result demonstrates that while colour can be exploited by colour-sensitive networks, it is not necessary to achieve high accuracy and so can be avoided if desired. We provide the public release of our statistical observational realism suite, RealSim, as a companion to this paper.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1093/mnras/stz2934

Cites

161 of 161 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Bottrell et al. (2019) model reached 87% accuracy.

    “Their model reached high (87%) accuracy even when using simulated images inserted into real SDSS survey fields to train and test the model.”

    From Detecting galaxy tidal features using self-supervised representation learning (2024) · cites this paper as Bottrell et al. (2019)

  • Both Bottrell et al. (2019) and Snyder et al. (2019) do not focus explicitly on the detection of tidal features and therefore are not directly comparable to our analysis.

    “However, both the works of Bottrell et al. (2019) and Snyder et al. (2019) do not focus explicitly on the detection of tidal features and therefore are not directly comparable to the results of our analysis.”

    From Detecting galaxy tidal features using self-supervised representation learning (2024) · cites this paper as Bottrell et al. (2019)

  • Bottrell et al. (2019) main conclusion is that it is more important that images used for training reproduce the observational properties of the sample to be analysed, that is, PSF, noise, background sources, than using full radiative transfer to improve the conversion from stellar particles to light.

    “The main conclusion is that it is more important that images used for training reproduce the observational properties of the sample to be analysed, that is, PSF, noise, background sources, than using full radiative transfer to improve the conversion from stellar particles to light.”

    From The Dawes Review 10: The impact of deep learning for the analysis of galaxy surveys (2023) · cites this paper as Bottrell et al. (2019)

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

4 of the 165 references couldn’t be matched to a known paper.

Research works printed but not linked: 9

  1. Barton E. J., Geller M. J., Kenyon S. J., 2000, The Astrophysical Journal, 530, 660
  2. Buda M., Maki A., Mazurowski M. A., 2017, arXiv e-prints, p. arXiv:1710.05381 · arXiv
  3. Cardie C., Howe N., 1997, in ICML. pp 57-65
  4. Chan P. K., Stolfo S. J., 1998, in KDD. pp 164-168
  5. Hausen R., Robertson B., 2019, arXiv e-prints, p. arXiv:1906.11248 · arXiv
  6. LeCun Y., Bottou L., Bengio Y., Haffner P., et al., 1998, Proceedings of the IEEE, 86, 2278
  7. Lupton R., Gunn J. E., Ivezi´ c Z., Knapp G. R., Kent S., 2001, in Harnden Jr. F. R., Primini F. A., Payne H. E., eds, Astronomical Society of the Pacific Conference Series Vol. 238, Astronomical Data Analysis Software and Systems X. p. 269 ( arXiv:astro-ph/0101420 ) · arXiv
  8. Pearson W. J., Wang L., Trayford J. W., Petrillo C. E., van der Tak F. F. S., 2019, A&A, 626, A49
  9. Radivojac P., Chawla N. V., Dunker A. K., Obradovic Z., 2004, Journal of Biomedical Informatics, 37, 224

Cited by

Sorted by citations

Related

Authors
Bottrell, C., Hani, M. H., Teimoorinia, H., Ellison, S. L., Moreno, J., Torrey, P., Hayward, C. C., Thorp, M., Simard, L., & Hernquist, L.
Affiliations
  • University of Victoria (CA)
  • California Institute of Technology (US)
  • University of Florida (US)
  • Flatiron Institute (US)
  • National Research Council Canada (CA)
  • Center for Astrophysics Harvard & Smithsonian (US)
Corresponding author
Bottrell, Connor
ORCID
Venue
Monthly Notices of the Royal Astronomical Society
Publisher
Oxford University Press (OUP)
Published
2019-10-18
Volume
490
Issue
4
Pages
5390-5413
ISSN
  • 0035-8711 (ISSN-L)
  • 1365-2966
DOI
10.1093/mnras/stz2934
OpenAlex
W2980735226
Cited by
113
References
165
Citations by year
2026: 5 · 2025: 16 · 2024: 22 · 2023: 21 · 2022: 26 · 2021: 9 · 2020: 12 · 2019: 2
FWCI
6.54
Type
journal article
Language
en
Open access
Licence
academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model (vor)
Locations
  • Monthly Notices of the Royal Astronomical Society (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • CaltechAUTHORS (California Institute of Technology) (submittedVersion)
  • NPARC (submittedVersion)
  • arXiv (Cornell University)
Funders
  • National Sciences and Engineering Research Council of Canada
  • NSF (1516374)
Topics
  • Physical Sciences › Physics and Astronomy › Astronomy and Astrophysics › Galaxies: Formation, Evolution, Phenomena
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Physics and Astronomy › Astronomy and Astrophysics › Astrophysical Phenomena and Observations
Keywords
galaxy mergers, synthetic galaxy images, radiative transfer, convolutional neural networks, point spread function, fluid simulation, galaxy morphology

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review