Hostname: page-component-8448b6f56d-dnltx Total loading time: 0 Render date: 2024-04-16T05:02:46.781Z Has data issue: false hasContentIssue false

Deep learning for studies of galaxy morphology

Published online by Cambridge University Press:  30 May 2017

D. Tuccillo
Affiliation:
GEPI, Observatoire de Paris, CNRS, Université Paris Diderot, 61, Avenue de l’Observatoire F-75014, Paris, France MINES ParisTech, PSL Research University, CMM Centre for mathematical morphology, Fontainebleau, France
M. Huertas-Company
Affiliation:
GEPI, Observatoire de Paris, CNRS, Université Paris Diderot, 61, Avenue de l’Observatoire F-75014, Paris, France
E. Decencière
Affiliation:
MINES ParisTech, PSL Research University, CMM Centre for mathematical morphology, Fontainebleau, France
S. Velasco-Forero
Affiliation:
MINES ParisTech, PSL Research University, CMM Centre for mathematical morphology, Fontainebleau, France
Rights & Permissions [Opens in a new window]

Abstract

Core share and HTML view are not available for this content. However, as you have access to this content, a full PDF is available via the ‘Save PDF’ action button.

Establishing accurate morphological measurements of galaxies in a reasonable amount of time for future big-data surveys such as EUCLID, the Large Synoptic Survey Telescope or the Wide Field Infrared Survey Telescope is a challenge. Because of its high level of abstraction with little human intervention, deep learning appears to be a promising approach. Deep learning is a rapidly growing discipline that models high-level patterns in data as complex multilayered networks. In this work we test the ability of deep convolutional networks to provide parametric properties of Hubble Space Telescope like galaxies (half-light radii, Sérsic indices, total flux etc..). We simulate a set of galaxies including point spread function and realistic noise from the CANDELS survey and try to recover the main galaxy parameters using deep-learning. We compare the results with the ones obtained with the commonly used profile fitting based software GALFIT. This way showing that with our method we obtain results at least equally good as the ones obtained with GALFIT but, once trained, with a factor 5 hundred time faster.

Type
Contributed Papers
Copyright
Copyright © International Astronomical Union 2017 

References

Bastien, F., et al., (2012), NIPS, 2012Google Scholar
Dieleman, S., Willett, K. W., & Dambre, J. 2015, MNRAS, 450, 1441 Google Scholar
Fukushima, K. 1980, Biological Cybernetics, 36, 193202 Google Scholar
Galametz, A., Grazian, A., Fontana, A., et al. 2013, ApJS, 206, 10 Google Scholar
Huertas-Company, M., et al. 2015, ApJS, 221, 8 CrossRefGoogle Scholar
Kim, E. & Brunner, R. 2017, MNRAS, 464, 44634475 Google Scholar
Koekemoer, A. M., et al. 2013, ApJS, 209, 3 CrossRefGoogle Scholar
Krizhevsky, A., Sutskever, I., & Hinton, G. 2012, NIPS1097–1105Google Scholar
LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., & Jackel, L. D. 1989, Neural Computation, 1 (4):541551 Google Scholar
Peng, C. Y., et al. 2002, AJ, 124, 266 Google Scholar
Schmidhuber, J. 2015, Neural Netw, 61, 85117 Google Scholar
van der Wel, A., et al. (2012) ApJS, 203, 24 CrossRefGoogle Scholar