Morphological classification of radio galaxies with Wasserstein generative adversarial network-supported augmentation

RAS Techniques and Instruments, vol. 2, pp. 264–277

Abstract

ABSTRACT Machine learning techniques that perform morphological classification of astronomical sources often suffer from a scarcity of labelled training data. Here, we focus on the case of supervised deep learning models for the morphological classification of radio galaxies, which is particularly topical for the forthcoming large radio surveys. We demonstrate the use of generative models, specifically Wasserstein generative adversarial networks (wGANs), to generate data for different classes of radio galaxies. Further, we study the impact of augmenting the training data with images from our wGAN on three different classification architectures. We find that this technique makes it possible to improve models for the morphological classification of radio galaxies. A simple fully connected neural network benefits most from including generated images into the training set, with a considerable improvement of its classification accuracy. In addition, we find it is more difficult to improve complex classifiers. The classification performance of a convolutional neural network can be improved slightly. However, this is not the case for a vision transformer.

Authors 10

1. Lennart Rustige corresponding

   Deutsches Elektronen-Synchrotron DESY

   Affiliation as printed

   Center for Data and Computing in Natural Sciences (CDCS) , Notkestrasse 9, D-22607 Hamburg, Germany

   Deutsches Elektronen-Synchrotron DESY , Notkestrasse 85, D-22607 Hamburg, Germany

2. Janis Kummer corresponding

   Universität Hamburg

   Affiliation as printed

   Center for Data and Computing in Natural Sciences (CDCS) , Notkestrasse 9, D-22607 Hamburg, Germany

   Universität Hamburg , Hamburger Sternwarte, Gojenbergsweg 112, D-21029 Hamburg, Germany

3. Florian Griese corresponding

   University Medical Center Hamburg-Eppendorf · Hamburg University of Technology

   Affiliation as printed

   Center for Data and Computing in Natural Sciences (CDCS) , Notkestrasse 9, D-22607 Hamburg, Germany

   Institute for Biomedical Imaging, Hamburg University of Technology , D-21073 Hamburg, Germany

   Section for Biomedical Imaging, University Medical Center Hamburg-Eppendorf , D-20246 Hamburg, Germany

4. K. Borras Aachen Physics Institute III A

   Deutsches Elektronen-Synchrotron DESY · RWTH Aachen University

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY , Notkestrasse 85, D-22607 Hamburg, Germany

   Physics Institute III A, RWTH Aachen University , Templergraben 55, D-52062 Aachen, Germany

5. M. Brüggen

   Universität Hamburg

   Affiliation as printed

   Universität Hamburg , Hamburger Sternwarte, Gojenbergsweg 112, D-21029 Hamburg, Germany

6. Patrick Connor

   Universität Hamburg

   Affiliation as printed

   Center for Data and Computing in Natural Sciences (CDCS) , Notkestrasse 9, D-22607 Hamburg, Germany

   Institut für Experimentalphysik, Universität Hamburg , Luruper Chaussee 149, D-22761 Hamburg, Germany

7. Frank Gaede

   Deutsches Elektronen-Synchrotron DESY

   Affiliation as printed

   Deutsches Elektronen-Synchrotron DESY , Notkestrasse 85, D-22607 Hamburg, Germany

8. Gregor Kasieczka

   Universität Hamburg

   Affiliation as printed

   Institut für Experimentalphysik, Universität Hamburg , Luruper Chaussee 149, D-22761 Hamburg, Germany

9. Tobias A. Knopp

   University Medical Center Hamburg-Eppendorf · Hamburg University of Technology

   Affiliation as printed

   Institute for Biomedical Imaging, Hamburg University of Technology , D-21073 Hamburg, Germany

   Section for Biomedical Imaging, University Medical Center Hamburg-Eppendorf , D-20246 Hamburg, Germany

10. Peter Schleper

    Universität Hamburg

    Affiliation as printed

    Institut für Experimentalphysik, Universität Hamburg , Luruper Chaussee 149, D-22761 Hamburg, Germany