CaltechAUTHORS
  A Caltech Library Service

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

Bottrell, Connor and Hani, Maan H. and Teimoorinia, Hossen and Ellison, Sara L. and Moreno, Jorge and Torrey, Paul and Hayward, Christopher C. and Thorp, Mallory and Simard, Luc and Hernquist, Lars (2019) Deep learning predictions of galaxy merger stage and the importance of observational realism. Monthly Notices of the Royal Astronomical Society, 490 (4). pp. 5390-5413. ISSN 0035-8711. https://resolver.caltech.edu/CaltechAUTHORS:20200109-143244301

[img] PDF - Published Version
See Usage Policy.

13Mb
[img] PDF - Accepted Version
See Usage Policy.

7Mb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20200109-143244301

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 per cent compared to 79.6 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 per cent with r-only compared to 87.1 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.


Item Type:Article
Related URLs:
URLURL TypeDescription
https://doi.org/10.1093/mnras/stz2934DOIArticle
https://arxiv.org/abs/1910.07031arXivDiscussion Paper
ORCID:
AuthorORCID
Bottrell, Connor0000-0003-4758-4501
Hani, Maan H.0000-0002-5351-2291
Ellison, Sara L.0000-0002-1768-1899
Torrey, Paul0000-0002-5653-0786
Hayward, Christopher C.0000-0003-4073-3236
Thorp, Mallory0000-0003-0080-8547
Hernquist, Lars0000-0001-6950-1629
Additional Information:© 2019 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) Accepted 2019 October 15. Received 2019 October 3; in original form 2019 August 21. CB acknowledges the support of a National Sciences and Engineering Research Council of Canada (NSERC) Graduate Scholarship. MHH and HT contributed equally to this research. MHH acknowledges the receipt of a Vanier Canada Graduate Scholarship. SLE and LS gratefully acknowledge support under the Canadian Discover Grants Programme. The data used in this paper were, in part, generated and hosted using facilities supported by the Scientific Computing Core at the Centre for Computational Astrophysics, a division of the Simons Foundation. The computations in this research were enabled in part by support provided by Compute Canada (www.computecanada.ca). The numerical simulations in this paper were run on the Odyssey cluster supported by the FAS Division of Science, Research Computing Group at Harvard University. Support for JM is provided by the NSF (AST Award Number 1516374), and by the Harvard Institute for Theory and Computation, through their Visiting Scholars Program. Funding for the SDSS IV has been provided by the Alfred P. Sloan Foundation, the U.S. Department of Energy Office of Science, and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High-Performance Computing at the University of Utah. The SDSS web site is www.sdss.org. SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, the Carnegie Institution for Science, Carnegie Mellon University, the Chilean Participation Group, the French Participation Group, Harvard-Smithsonian Center for Astrophysics, Instituto de Astrofísica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU)/University of Tokyo, the Korean Participation Group, Lawrence Berkeley National Laboratory, Leibniz Institut für Astrophysik Potsdam (AIP), Max-Planck-Institut für Astronomie (MPIA Heidelberg), Max-Planck-Institut für Astrophysik (MPA Garching), Max-Planck-Institut für Extraterrestrische Physik (MPE), National Astronomical Observatories of China, New Mexico State University, New York University, University of Notre Dame, Observatário Nacional/MCTI, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group, Universidad Nacional Autónoma de México, University of Arizona, University of Colorado Boulder, University of Oxford, University of Portsmouth, University of Utah, University of Virginia, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University.
Group:TAPIR
Funders:
Funding AgencyGrant Number
Natural Sciences and Engineering Research Council of Canada (NSERC)UNSPECIFIED
Vanier Canada Graduate ScholarshipUNSPECIFIED
Simons FoundationUNSPECIFIED
Compute CanadaUNSPECIFIED
NSFAST-1516374
Harvard Institute for Theory and ComputationUNSPECIFIED
Subject Keywords:Methods: data analysis, Methods: numerical, Techniques: image processing, Galaxies: general, Galaxies: interactions, Galaxies: photometry
Issue or Number:4
Record Number:CaltechAUTHORS:20200109-143244301
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20200109-143244301
Official Citation:Connor Bottrell, Maan H Hani, Hossen Teimoorinia, Sara L Ellison, Jorge Moreno, Paul Torrey, Christopher C Hayward, Mallory Thorp, Luc Simard, Lars Hernquist, Deep learning predictions of galaxy merger stage and the importance of observational realism, Monthly Notices of the Royal Astronomical Society, Volume 490, Issue 4, December 2019, Pages 5390–5413, https://doi.org/10.1093/mnras/stz2934
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:100613
Collection:CaltechAUTHORS
Deposited By: George Porter
Deposited On:10 Jan 2020 16:38
Last Modified:10 Jan 2020 16:38

Repository Staff Only: item control page