Radio galaxy zoo: Knowledge transfer using rotationally invariant self-organizing maps

T. J. Galvin, M. Huynh, R. P. Norris, X. R. Wang, E. Hopkins, O. I. Wong, S. Shabala, L. Rudnick, M. J. Alger, K. L. Polsterer

Research output: Contribution to journalArticle

2 Citations (Scopus)
9 Downloads (Pure)

Abstract

With the advent of large scale-surveys the manual analysis and classification of individual radio source morphologies is rendered impossible as existing approaches do not scale. The analysis of complex morphological features in the spatial domain is a particularly important task. Here, we discuss the challenges of transferring crowdsourced labels obtained from the Radio Galaxy Zoo project and introduce a proper transfer mechanism via quantile random forest regression. By using parallelized rotation and flipping invariant Kohonen-maps, image cubes of Radio Galaxy Zoo selected galaxies formed from the Faint Images of the Radio Sky at Twenty-cm (FIRST) radio continuum and the Wide-field Infrared Survey Explorer (WISE) infrared all-sky surveys are first projected down to a two-dimensional embedding in an unsupervised way. This embedding can be seen as a discretized space of shapes with the coordinates reflecting morphological features as expressed by the automatically derived prototypes. We find that these prototypes have reconstructed physically meaningful processes across two channel images at radio and infrared wavelengths in an unsupervised manner. In the second step, images are compared with those prototypes to create a heat map, which is the morphological fingerprint of each object and the basis for transferring the user generated labels. These heat maps have reduced the feature space by a factor of 248, and are able to be used as the basis for subsequent machine-learning (ML) methods. Using an ensemble of decision trees we achieve upwards of 85.7% and 80.7% accuracy when predicting the number of components and peaks in an image, respectively, using these heat maps. We also question the currently used discrete classification schema and introduce a continuous scale that better reflects the uncertainty in transition between two classes, caused by sensitivity and resolution limits.

Original languageEnglish
Article number108009
JournalPublications of the Astronomical Society of the Pacific
Volume131
Issue number1004
DOIs
Publication statusPublished - 1 Oct 2019

Fingerprint

zoo
organizing
radio galaxies
radio
prototypes
heat
embedding
quantiles
Wide-field Infrared Survey Explorer
machine learning
sky
regression analysis
galaxies
continuums
sensitivity
wavelength
wavelengths

Cite this

Galvin, T. J. ; Huynh, M. ; Norris, R. P. ; Wang, X. R. ; Hopkins, E. ; Wong, O. I. ; Shabala, S. ; Rudnick, L. ; Alger, M. J. ; Polsterer, K. L. / Radio galaxy zoo : Knowledge transfer using rotationally invariant self-organizing maps. In: Publications of the Astronomical Society of the Pacific. 2019 ; Vol. 131, No. 1004.
@article{06588211d9744a1d97d30aecb060cfa8,
title = "Radio galaxy zoo: Knowledge transfer using rotationally invariant self-organizing maps",
abstract = "With the advent of large scale-surveys the manual analysis and classification of individual radio source morphologies is rendered impossible as existing approaches do not scale. The analysis of complex morphological features in the spatial domain is a particularly important task. Here, we discuss the challenges of transferring crowdsourced labels obtained from the Radio Galaxy Zoo project and introduce a proper transfer mechanism via quantile random forest regression. By using parallelized rotation and flipping invariant Kohonen-maps, image cubes of Radio Galaxy Zoo selected galaxies formed from the Faint Images of the Radio Sky at Twenty-cm (FIRST) radio continuum and the Wide-field Infrared Survey Explorer (WISE) infrared all-sky surveys are first projected down to a two-dimensional embedding in an unsupervised way. This embedding can be seen as a discretized space of shapes with the coordinates reflecting morphological features as expressed by the automatically derived prototypes. We find that these prototypes have reconstructed physically meaningful processes across two channel images at radio and infrared wavelengths in an unsupervised manner. In the second step, images are compared with those prototypes to create a heat map, which is the morphological fingerprint of each object and the basis for transferring the user generated labels. These heat maps have reduced the feature space by a factor of 248, and are able to be used as the basis for subsequent machine-learning (ML) methods. Using an ensemble of decision trees we achieve upwards of 85.7{\%} and 80.7{\%} accuracy when predicting the number of components and peaks in an image, respectively, using these heat maps. We also question the currently used discrete classification schema and introduce a continuous scale that better reflects the uncertainty in transition between two classes, caused by sensitivity and resolution limits.",
keywords = "Galaxies: general, Galaxies: jets, Galaxies: statistics, Infrared: general, Radio continuum: general",
author = "Galvin, {T. J.} and M. Huynh and Norris, {R. P.} and Wang, {X. R.} and E. Hopkins and Wong, {O. I.} and S. Shabala and L. Rudnick and Alger, {M. J.} and Polsterer, {K. L.}",
year = "2019",
month = "10",
day = "1",
doi = "10.1088/1538-3873/ab150b",
language = "English",
volume = "131",
journal = "Publications of the Astronomical Society of the Pacific",
issn = "0004-6280",
publisher = "University of Chicago",
number = "1004",

}

Galvin, TJ, Huynh, M, Norris, RP, Wang, XR, Hopkins, E, Wong, OI, Shabala, S, Rudnick, L, Alger, MJ & Polsterer, KL 2019, 'Radio galaxy zoo: Knowledge transfer using rotationally invariant self-organizing maps' Publications of the Astronomical Society of the Pacific, vol. 131, no. 1004, 108009. https://doi.org/10.1088/1538-3873/ab150b

Radio galaxy zoo : Knowledge transfer using rotationally invariant self-organizing maps. / Galvin, T. J.; Huynh, M.; Norris, R. P.; Wang, X. R.; Hopkins, E.; Wong, O. I.; Shabala, S.; Rudnick, L.; Alger, M. J.; Polsterer, K. L.

In: Publications of the Astronomical Society of the Pacific, Vol. 131, No. 1004, 108009, 01.10.2019.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Radio galaxy zoo

T2 - Knowledge transfer using rotationally invariant self-organizing maps

AU - Galvin, T. J.

AU - Huynh, M.

AU - Norris, R. P.

AU - Wang, X. R.

AU - Hopkins, E.

AU - Wong, O. I.

AU - Shabala, S.

AU - Rudnick, L.

AU - Alger, M. J.

AU - Polsterer, K. L.

PY - 2019/10/1

Y1 - 2019/10/1

N2 - With the advent of large scale-surveys the manual analysis and classification of individual radio source morphologies is rendered impossible as existing approaches do not scale. The analysis of complex morphological features in the spatial domain is a particularly important task. Here, we discuss the challenges of transferring crowdsourced labels obtained from the Radio Galaxy Zoo project and introduce a proper transfer mechanism via quantile random forest regression. By using parallelized rotation and flipping invariant Kohonen-maps, image cubes of Radio Galaxy Zoo selected galaxies formed from the Faint Images of the Radio Sky at Twenty-cm (FIRST) radio continuum and the Wide-field Infrared Survey Explorer (WISE) infrared all-sky surveys are first projected down to a two-dimensional embedding in an unsupervised way. This embedding can be seen as a discretized space of shapes with the coordinates reflecting morphological features as expressed by the automatically derived prototypes. We find that these prototypes have reconstructed physically meaningful processes across two channel images at radio and infrared wavelengths in an unsupervised manner. In the second step, images are compared with those prototypes to create a heat map, which is the morphological fingerprint of each object and the basis for transferring the user generated labels. These heat maps have reduced the feature space by a factor of 248, and are able to be used as the basis for subsequent machine-learning (ML) methods. Using an ensemble of decision trees we achieve upwards of 85.7% and 80.7% accuracy when predicting the number of components and peaks in an image, respectively, using these heat maps. We also question the currently used discrete classification schema and introduce a continuous scale that better reflects the uncertainty in transition between two classes, caused by sensitivity and resolution limits.

AB - With the advent of large scale-surveys the manual analysis and classification of individual radio source morphologies is rendered impossible as existing approaches do not scale. The analysis of complex morphological features in the spatial domain is a particularly important task. Here, we discuss the challenges of transferring crowdsourced labels obtained from the Radio Galaxy Zoo project and introduce a proper transfer mechanism via quantile random forest regression. By using parallelized rotation and flipping invariant Kohonen-maps, image cubes of Radio Galaxy Zoo selected galaxies formed from the Faint Images of the Radio Sky at Twenty-cm (FIRST) radio continuum and the Wide-field Infrared Survey Explorer (WISE) infrared all-sky surveys are first projected down to a two-dimensional embedding in an unsupervised way. This embedding can be seen as a discretized space of shapes with the coordinates reflecting morphological features as expressed by the automatically derived prototypes. We find that these prototypes have reconstructed physically meaningful processes across two channel images at radio and infrared wavelengths in an unsupervised manner. In the second step, images are compared with those prototypes to create a heat map, which is the morphological fingerprint of each object and the basis for transferring the user generated labels. These heat maps have reduced the feature space by a factor of 248, and are able to be used as the basis for subsequent machine-learning (ML) methods. Using an ensemble of decision trees we achieve upwards of 85.7% and 80.7% accuracy when predicting the number of components and peaks in an image, respectively, using these heat maps. We also question the currently used discrete classification schema and introduce a continuous scale that better reflects the uncertainty in transition between two classes, caused by sensitivity and resolution limits.

KW - Galaxies: general

KW - Galaxies: jets

KW - Galaxies: statistics

KW - Infrared: general

KW - Radio continuum: general

UR - http://www.scopus.com/inward/record.url?scp=85073532371&partnerID=8YFLogxK

U2 - 10.1088/1538-3873/ab150b

DO - 10.1088/1538-3873/ab150b

M3 - Article

VL - 131

JO - Publications of the Astronomical Society of the Pacific

JF - Publications of the Astronomical Society of the Pacific

SN - 0004-6280

IS - 1004

M1 - 108009

ER -