J/A+A/659/A144 J-PLUS. STAR-GALAXY-QSO Classification (Wang+, 2022)
J-PLUS: Support vector machine applied to STAR-GALAXY-QSO classification.
Wang C., Bai Y., Lopez-Sanjuan C., Yuan H., Wang S., Liu J., Sobral D.,
Baqui P.O., Martin E.L., Galarza C.A., Alcaniz J., Angulo R.E.,
Cenarro A.J., Cristobal-Hornillos D., Dupke R.A., Ederoclite A.,
Hernandez-Monteagudo C., Marin-Franch A., Moles M., Sodre L. Jr,
Vazquez Ramio H., Varela J.
<Astron. Astrophys. 659, A144 (2022)>
=2022A&A...659A.144W 2022A&A...659A.144W (SIMBAD/NED BibCode)
ADC_Keywords: Surveys ; Photometry
Keywords: methods: data analysis - techniques: spectroscopic -
astronomical databases: miscellaneous
Abstract:
In modern astronomy, machine learning has proved to be efficient and
effective in mining big data from the newest telescopes.
In this study, we construct a supervised machine-learning algorithm to
classify the objects in the Javalambre Photometric Local Universe
Survey first data release (J-PLUS DR1).
The sample set is featured with 12-waveband photometry and labeled
with spectrum-based catalogs, including Sloan Digital Sky Survey
(SDSS) spectroscopic data, the Large Sky Area Multi-Object Fiber
Spectroscopic Telescope (LAMOST), and VERON- CAT - the Veron Catalog
of Quasars & AGN (VV13. Cat. VII/258). The performance of the
classifier is presented with the applications of blind test
validations based on RAdial Velocity Extension (RAVE), the Kepler
Input Catalog (KIC), the 2 MASS (the Two Micron All Sky Survey)
Redshift Survey (2MRS), and the UV-bright Quasar Survey (UVQS). A new
algorithm was applied to constrain the potential extrapolation that
could decrease the performance of the machine-learning classifier.
The accuracies of the classifier are 96.5% in the blind test and 97.0%
in training cross-validation. The F1-scores for each class are
presented to show the balance between the precision and the recall of
the classifier. We also discuss different methods to constrain the
potential extrapolation.
Description:
The machine-learning sample is made up of SDSS, LAMOST, and VV13
(Table 1, see more in Appendix C, and magnitude distributions are in
Appendix B). There are 468685 unique objects with 12 valid
magnitudes, including 74701 galaxies, 45899 QSOs, and 348085 stars.
These 468 685 objects were all put in training with a 10-fold
validation.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
tablec1.dat 422 468685 Sample set
table6.dat 92 3496867 Interpolation set
table7.dat 93 630061 Extrapolation set
table9.dat 93 155 Ambiguous set
table10.dat 178 26 Abnormal set
--------------------------------------------------------------------------------
Byte-by-byte Description of file: tablec1.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 11 A11 --- J-PLUS J-PLUS designation
13- 23 E11.9 deg RAdeg Right ascension (J2000.0)
25- 36 F12.9 deg DEdeg Declination (J2000.0)
38- 43 A6 --- Class Cross matched class
45- 54 A10 --- subclass Cross matched subclass for stars.
Blank means not found or not a star
56- 73 A18 --- Cat Catalog whom provides this object
75- 86 F12.9 mag umag u band magnitude (1)
88- 99 F12.9 mag J0378mag J0378 magnitude (1)
101-112 F12.9 mag J0395mag J0395 magnitude (1)
114-125 F12.9 mag J0410mag J0410 magnitude (1)
127-138 F12.9 mag J0430mag J0430 magnitude (1)
140-151 F12.9 mag gmag g band magnitude (1)
153-164 F12.9 mag J0515mag J0515 magnitude (1)
166-177 F12.9 mag rmag r band magnitude (1)
179-190 F12.9 mag J0660mag J0660 magnitude (1)
192-203 F12.9 mag imag i band magnitude (1)
205-216 F12.9 mag J0861mag J0861 magnitude (1)
218-229 F12.9 mag zmag z band magnitude (1)
231-242 F12.9 mag e_umag u band magnitude error (1)
244-257 F14.9 mag e_J0378mag J0378 magnitude error (1)
259-272 F14.9 mag e_J0395mag J0395 magnitude error (1)
274-287 F14.9 mag e_J0410mag J0410 magnitude error (1)
289-302 F14.9 mag e_J0430mag J0430 magnitude error (1)
304-318 F15.9 mag e_gmag g band magnitude error (1)
320-333 F14.9 mag e_J0515mag J0515 magnitude error (1)
335-349 F15.9 mag e_rmag r band magnitude error (1)
351-365 F15.9 mag e_J0660mag J0660 magnitude error (1)
367-380 F14.9 mag e_imag i band magnitude error (1)
382-394 F13.9 mag e_J0861mag J0861 magnitude error (1)
396-409 F14.9 mag e_zmag z band magnitude error (1)
411-422 F12.9 --- Weight Training weight
--------------------------------------------------------------------------------
Note (1): All magnitudes and uncertainties are given by Haibo Yuan's
recalibrated J-PLUS catalog.
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table6.dat table7.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 11 A11 --- J-PLUS J-PLUS designation
13- 31 F19.15 deg RAdeg Right ascension (J2000)
33- 46 F14.11 deg DEdeg Declination (J2000)
48- 67 E20.18 --- ClassS J-PLUS parameter class_star. The probability
of the object is a star given by J-PLUS
69- 74 A6 --- PClass Label given from our classification
76- 93 F18.16 --- Prob Probability of the object belongs to the
label in PredictClass given by our method
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table9.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 11 A11 --- J-PLUS J-PLUS designation
13- 25 F13.9 deg RAdeg Right ascension (J2000.0)
27- 38 F12.9 deg DEdeg Declination (J2000)
40- 50 E11.9 --- ClassS J-PLUS parameter class_star. The probability
of the object is a star given by J-PLUS.
52- 57 A6 --- PClass Label given from our classification
59- 69 F11.9 --- Galaxy Probability of the object is a galaxy
given by our method
71- 81 F11.9 --- QSO Probability of the object is a QSO
given by our method
83- 93 F11.9 --- Star Probability of the object is a star
given by our method
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table10.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 11 A11 --- J-PLUS J-PLUS designation
13- 27 F15.11 deg RAdeg Right ascension (J2000.0)
29- 44 F16.11 deg DEdeg Declination (J2000.0)
46- 65 E20.17 --- ClassS J-PLUS parameter class_star. The probability
of the object is a star given by J-PLUS
67- 72 A6 --- PClass Label given from our classification
74- 90 F17.15 --- Galaxy Probability of the object is a galaxy
given by our method
92-108 F17.13 --- Gdist Mahalanobis distance of the object to the
group galaxy
110-126 F17.15 --- QSO Probability of the object is a QSO
given by our method
128-143 F16.13 --- Qdis Mahalanobis distance of the object to the
group QSO
145-161 F17.15 --- Star Probability of the object is a star
given by our method
163-178 F16.12 --- Sdis Mahalanobis distance of the object to the
group star
--------------------------------------------------------------------------------
Acknowledgements:
Cunshi Wang, wangcunshi(at)nao.cas.cn
(End) Patricia Vannier [CDS] 24-Jan-2022