J/MNRAS/479/415 Galaxy mergers sample (Ackermann+, 2018)
Using transfer learning to detect galaxy mergers.
Ackermann S., Schawinski K., Zhang C., Weigel A.K., Turp M.D.
<Mon. Not. R. Astron. Soc., 479, 415-425 (2018)>
=2018MNRAS.479..415A 2018MNRAS.479..415A (SIMBAD/NED BibCode)
ADC_Keywords: Galaxy catalogs
Keywords: methods: data analysis - techniques: image processing -
galaxies: general
Abstract:
We investigate the use of deep convolutional neural networks (deep
CNNs) for automatic visual detection of galaxy mergers. Moreover, we
investigate the use of transfer learning in conjunction with CNNs by
retraining networks first trained on pictures of everyday objects. We
test the hypothesis that transfer learning is useful for improving
classification performance for small training sets. This would make
transfer learning useful for finding rare objects in astronomical
imaging data sets. We find that these deep learning methods perform
significantly better than current state-of-the-art merger detection
methods based on non-parametric systems such as CAS and GM20. Our
method is end-to-end and robust to image noise and distortions; it can
be applied directly without image preprocessing. We also find that
transfer learning can act as a regularizer in some cases, leading to
better overall classification accuracy (p=0.02). Transfer learning
on our full training set leads to a lowered error rate from 0.038±1
to 0.032±1, a relative improvement of 15 per cent. Finally, we
perform a basic sanity-check by creating a merger sample with our
method, and comparing with an already existing, manually created
merger catalogue in terms of colour--mass distribution and stellar
mass function.
Description:
We created a merger sample by taking all the GZ I (Galaxy Zoo I)
objects in the same redshift range as the Darg et al.
(2010MNRAS.401.1043D 2010MNRAS.401.1043D) merger catalogue, and then obtaining the
classification pm for each galaxy with our classifier. We used the
classifier trained on the largest training set with transfer learning.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table2.dat 134 328151 Merger sample
--------------------------------------------------------------------------------
See also:
http://zoo1.galaxyzoo.org : Galaxy Zoo Home Page
Byte-by-byte Description of file: table2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 I18 --- objID SDSS objID
20- 21 I2 h RAh Right ascension (J2000)
23- 24 I2 min RAm Right ascension (J2000)
26- 30 F5.2 s RAs Right ascension (J2000)
32 A1 --- DE- Declination sign (J2000)
33- 34 I2 deg DEd Declination (J2000)
36- 37 I2 arcmin DEm Declination (J2000)
39- 42 F4.1 arcsec DEs Declination (J2000)
44- 65 E22.20 --- pmerger0 [0/1]? Merger probability estimate
67- 88 E22.20 --- pmerger1 [0/1]? Merger probability estimate
90-111 E22.20 --- pmerger2 [0/1]? Merger probability estimate
113-134 E22.20 --- pmerger3 [0/1]? Merger probability estimate
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Patricia Vannier [CDS] 28-Apr-2022