AutoSource: Radio-astronomical source-finding with convolutional autoencoders. (arXiv:1910.03631v1 [astro-ph.IM])
<a href="http://arxiv.org/find/astro-ph/1/au:+Lukic_V/0/1/0/all/0/1">V. Lukic</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Gasperin_F/0/1/0/all/0/1">F. De Gasperin</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Bruggen_M/0/1/0/all/0/1">M. Br&#xfc;ggen</a>

Finding and classifying astronomical sources is key in the scientific
exploitation of radio surveys. Source-finding usually involves identifying the
parts of an image belonging to an astronomical source, against some estimated
background. This can be problematic in the radio regime, owing to the presence
of correlated noise, which can interfere with the signal from the source. In
the current work we present AutoSource, a novel method based on a deep learning
technique, to identify the positions of radio sources, and compare the results
to a Gaussian-fitting method. Since the deep learning approach allows the
generation of more training images, it should perform well in the
source-finding task. We test the source-finding methods on artificial data
created for the data challenge of the Square Kilometre Array (SKA). We
investigate sources that are divided into three classes: star-forming galaxies
(SFGs) and two classes of Active Galactic Nuclei (AGN). The artifical data is
given at two different frequencies (560 MHz and 1400 MHz), three total
integration times (8 h, 100 h, 1000 h) and three signal-to-noise ratios (1, 2,
and 5). At lower SNRs, AutoSource tends to outperform a Gaussian-fitting
approach in the recovery of SFGs and all sources, although at the lowest SNR of
1, the better performance is likely due to chance matches. The Gaussian-fitting
method performs better in the recovery of the AGN-type sources at lower SNRs.
At a higher SNR, AutoSource performs better on average in the recovery of AGN
sources, whereas the Gaussian-fitting method performs better in the recovery of
SFGs and all sources. AutoSource usually performs better at shorter total
integration times, and detects more true positives and misses fewer sources
compared to the Gaussian-fitting method, however it detects more false
positives.

Finding and classifying astronomical sources is key in the scientific
exploitation of radio surveys. Source-finding usually involves identifying the
parts of an image belonging to an astronomical source, against some estimated
background. This can be problematic in the radio regime, owing to the presence
of correlated noise, which can interfere with the signal from the source. In
the current work we present AutoSource, a novel method based on a deep learning
technique, to identify the positions of radio sources, and compare the results
to a Gaussian-fitting method. Since the deep learning approach allows the
generation of more training images, it should perform well in the
source-finding task. We test the source-finding methods on artificial data
created for the data challenge of the Square Kilometre Array (SKA). We
investigate sources that are divided into three classes: star-forming galaxies
(SFGs) and two classes of Active Galactic Nuclei (AGN). The artifical data is
given at two different frequencies (560 MHz and 1400 MHz), three total
integration times (8 h, 100 h, 1000 h) and three signal-to-noise ratios (1, 2,
and 5). At lower SNRs, AutoSource tends to outperform a Gaussian-fitting
approach in the recovery of SFGs and all sources, although at the lowest SNR of
1, the better performance is likely due to chance matches. The Gaussian-fitting
method performs better in the recovery of the AGN-type sources at lower SNRs.
At a higher SNR, AutoSource performs better on average in the recovery of AGN
sources, whereas the Gaussian-fitting method performs better in the recovery of
SFGs and all sources. AutoSource usually performs better at shorter total
integration times, and detects more true positives and misses fewer sources
compared to the Gaussian-fitting method, however it detects more false
positives.

http://arxiv.org/icons/sfx.gif