J/A+A/611/A97 Photometric quasar candidates in Stripe 82 (Pasquet-Itam+, 2018)
Deep learning approach for classifying, detecting and predicting photometric
redshifts of quasars in the Sloan Digital Sky Survey stripe 82.
Pasquet-Itam J., Pasquet J.
<Astron. Astrophys. 611, A97 (2018)>
=2018A&A...611A..97P 2018A&A...611A..97P (SIMBAD/NED BibCode)
ADC_Keywords: QSOs ; Photometry, SDSS
Keywords: methods: data analysis - techniques: photometric -
techniques: image processing - quasars: general - surveys
Abstract:
We have applied a convolutional neural network (CNN) to classify and
detect quasars in the Sloan Digital Sky Survey Stripe 82 and also to
predict the photometric redshifts of quasars. The network takes the
variability of objects into account by converting light curves into
images. The width of the images, noted w, corresponds to the five
magnitudes ugriz and the height of the images, noted h, represents the
date of the observation. The CNN provides good results since its
precision is 0.988 for a recall of 0.90, compared to a precision of
0.985 for the same recall with a random forest classifier. Moreover
175 new quasar candidates are found with the CNN considering a fixed
recall of 0.97. The combination of probabilities given by the CNN and
the random forest makes good performance even better with a precision
of 0.99 for a recall of 0.90. For the redshift predictions, the CNN
presents excellent results which are higher than those obtained with a
feature extraction step and different classifiers
(a K-nearest-neighbors, a support vector machine, a random forest and
a Gaussian process classifier). Indeed, the accuracy of the CNN within
|{DELTA}z|<0.1 can reach 78.09%, within |{DELTA}z|<0.2 reaches 86.15%,
within |{DELTA}z|<0.3 reaches 91.2% and the value of root mean square
(rms) is 0.359. The performance of the KNN decreases for the three
|{DELTA}z| regions, since within the accuracy of |{DELTA}z|<0.1,
|{DELTA}z|<0.2, and |{DELTA}z|<0.3 is 73.72%, 82.46%, and 90.09%
respectively, and the value of rms amounts to 0.395. So the CNN
successfully reduces the dispersion and the catastrophic redshifts of
quasars. This new method is very promising for the future of big
databases such as the Large Synoptic Survey Telescope.
Description:
We present a list of 175 new quasar candidates detected by a
convolutional neural network (CNN) in the Sloan Digital Sky Survey
Stripe 82 with a fixed recall of 0.97. The imaging data used in our
work consists of objects solely from the publicly available variable
source catalog (UWVSC; Ivezic et al. (2007, Cat. J/AJ/134/973), Sesar
et al. (2007AJ....134.2236S 2007AJ....134.2236S)) constructed by researchers at the
University of Washington. This catalog contains 67,507 unresolved,
variable candidates with g≤20.5mag, at least 10 observations in
both g and r bands, and a light curve with a root-mean-scatter
(rms)>0.05mag and chi2 per degree of freedom >3 in both g and r
bands. The CNN takes the variability of objects into account by
converting light curves into images. The CNN provides good results
since its precision is 0.988 for a recall of 0.90, compared to a
precision of 0.985 for the same recall with a random forest
classifier.The combination of probabilities given by the CNN and the
random forest makes good performance even better with a precision of
0.99 for a recall of 0.90. For each a 7-character long integer ID, the
candidate equatorial coordinates (decimal degrees, J2000), a flag
which indicates the label given by the random forest method
(0 = quasar, 1 = other),the r-band magnitude (corrected for ISM
extinction) and the u-g, g-r, r-i, i-z SDSS colors (corrected for ISM
extinction) are given.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
qso.dat 61 175 Quasar candidates
--------------------------------------------------------------------------------
See also:
J/AJ/134/973 : SDSS Stripe 82 star catalogs (Ivezic+, 2007)
Byte-by-byte Description of file: qso.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 7 I7 --- ID Stripe 82 identification number
9- 18 F10.6 deg RAdeg Right ascension (J2000.0)
20- 29 F10.6 deg DEdeg Declination (J2000.0)
31 A1 --- RF [0/1] label given by the Random Forest
(0=quasar, 1=other) (1)
33- 37 F5.2 mag rmag SDSS r-median magnitude (2)
39- 43 F5.2 mag u-g SDSS u-g color index (2)
45- 49 F5.2 mag g-r SDSS g-r color index (2)
51- 55 F5.2 mag r-i SDSS r-i color index (2)
57- 61 F5.2 mag i-z SDSS i-z color index (2)
--------------------------------------------------------------------------------
Note (1): label as follows:
0 = quasar
1 = other
Note (2): corrected for ISM extinction.
--------------------------------------------------------------------------------
Acknowledgements:
Johanna Pasquet Itam, pasquet(at)cppm.in2p3.fr
(End) Patricia Vannier [CDS] 14-Nov-2017