J/A+A/691/A221 J-PLUS Bayesian object classification (del Pino+, 2024)
J-PLUS: Bayesian object classification with a strum of BANNJOS.
del Pino A., Lopez-Sanjuan C., Hernan-Caballero A., Dominguez-Sanchez H.,
von Marttens R., Fernandez-Ontiveros J.A., Coelho P.R.T.,
Lumbreras-Calle A., Vega-Ferrero J., Jimenez-Esteban F., Cruz P., Marra V.,
Quartin M., Galarza C.A., 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, Varela J., Vazquez Ramio H.
<Astron. Astrophys. 691, A221 (2024)>
=2024A&A...691A.221D 2024A&A...691A.221D (SIMBAD/NED BibCode)
ADC_Keywords: Surveys ; Galaxies ; QSOs ; Photometry
Keywords: methods: data analysis - catalogs - Galaxy: stellar content -
quasars: general - galaxies: statistics
Abstract:
With its 12 optical filters, the Javalambre-Photometric Local Universe
Survey (J-PLUS) provides an unprecedented multicolor view of the local
Universe. The third data release (DR3) covers 3,192 deg^2 and contains
47.4 million objects. However, the classification algorithms currently
implemented in the J-PLUS pipeline are deterministic and based solely
on the sources morphology of the sources.
Our goal is to classify the sources identified in the J-PLUS DR3
images into stars, quasi-stellar objects (QSOs), and galaxies. For
this task, we present BANNJOS, a machine learning pipeline that
utilizes Bayesian neural networks to provide the full probability
distribution function (PDF) of the classification.
BANNJOS is trained on photometric, astrometric, and morphological data
from J-PLUS DR3, Gaia DR3, and Catwise, using over 1.2 million objects
with spectroscopic classification from SDSS DR18, LAMOST DR9, DESI
Early Data Release, and Gaia DR3. Results are validated on a test set
of about 140,000 objects and cross-checked against theoretical model
predictions.
BANNJOS outperforms all previous classifiers in terms of accuracy,
precision, and completeness across the entire magnitude range. It
delivers over 95% accuracy for objects brighter than r=21.5 mag, and
∼90% accuracy for those up to r=22mag, where J-PLUS completeness is
<25%. BANNJOS is also the first object classifier to provide the full
probability distribution function (PDF) of the classification,
enabling precise object selection for high purity or completeness, and
for identifying objects with complex features, like active galactic
nuclei with resolved host galaxies.
BANNJOS has effectively classified J-PLUS sources into around 20
million galaxies, 1 million QSOs, and 26 million stars, with full PDFs
for each, which allow for later refinement of the sample. The upcoming
J-PAS survey, with its 56 color bands, will further enhance BANNJOS's
ability to detail the nature of each source.
Description:
The catalog contains Bayesian classification probabilities for stars,
QSOs, and galaxies, along with various statistical measures of the
classification. Each row corresponds to an object detected in the
J-PLUS DR3 images. The table also provides covariance terms and
Gaussian component details for classification probability
distributions.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
catalog.dat 449 47793197 Bayesian classification data (table F1)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: catalog.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 10 E10.6 deg RAdeg Right ascension (J2000) (alpha_j2000)
12- 20 E9.6 deg DEdeg Declination (J2000) (delta_j2000)
22- 27 I6 --- TileId Identifier of the Tile (tile_id)
29- 34 I6 --- IdNumber Number assigned by SExtractor (number)
36- 41 F6.3 mag rmag ?=99 Magnitude AB in r band
(magautorSDSSB)
43- 51 E9.6 --- ClassGal Mean of galaxy probability
(CLASSGALAXYmean)
53- 61 E9.6 --- ClassQSO Mean of QSO probability
(CLASSQSOmean)
63- 71 E9.6 --- ClassStar Mean of star probability
(CLASSSTARmean)
73- 80 F8.6 --- e_ClassGal MAD standard deviation of galaxy
(CLASSGALAXYstd)
82- 89 F8.6 --- e_ClassQSO MAD standard deviation of QSO
(CLASSQSOstd)
91- 98 F8.6 --- e_ClassStar MAD standard deviation of star
(CLASSSTARstd)
100-108 E9.6 --- ClassGal-QSOcorr Correlation between galaxy-QSO
(CLASSGALAXYCLASSQSOcorr)
110-118 E9.6 --- ClassGAL-Starcorr Correlation between galaxy-star
(CLASSGALAXYCLASSSTARcorr)
120-128 E9.6 --- ClassQSO-Starcorr Correlation between QSO-star
(CLASSQSOCLASSSTARcorr)
130-138 E9.6 --- ClassGal02 2.275th percentile of galaxy
(CLASSGALAXYpc02)
140-148 E9.6 --- ClassQSO02 2.275th percentile of QSO
(CLASSQSOpc02)
150-158 E9.6 --- ClassStar02 2.275th percentile of star
(CLASSSTARpc02)
160-168 E9.6 --- ClassGal16 15.865th percentile of galaxy
(CLASSGALAXYpc16)
170-178 E9.6 --- ClassQSO16 15.865th percentile of QSO
(CLASSQSOpc16)
180-188 E9.6 --- ClassStar16 15.865th percentile of star
(CLASSSTARpc16)
190-198 E9.6 --- ClassGal50 Median of galaxy probability
(CLASSGALAXYpc50)
200-208 E9.6 --- ClassQSO50 Median of QSO probability
(CLASSQSOpc50)
210-218 E9.6 --- ClassStar50 Median of star probability
(CLASSSTARpc50)
220-228 F9.6 --- ClassGal84 84.135th percentile of galaxy
(CLASSGALAXYpc84)
230-238 F9.6 --- ClassQSO84 84.135th percentile of QSO
(CLASSQSOpc84)
240-248 E9.6 --- ClassStar84 84.135th percentile of star
(CLASSSTARpc84)
250-258 F9.6 --- ClassGal98 97.725th percentile of galaxy
(CLASSGALAXYpc98)
260-268 F9.6 --- ClassQSO98 97.725th percentile of QSO
(CLASSQSOpc98)
270-278 F9.6 --- ClassStar98 97.725th percentile of star
(CLASSSTARpc98)
280-287 E8.6 --- comp1Cov11 Covariance term 11 for component 1
(comp1cov11)
289-297 E9.6 --- comp1Cov12 Covariance term 12 for component 1
(comp1cov12)
299-306 E8.6 --- comp1Cov22 Covariance term 22 for component 1
(comp1cov22)
308-315 E8.6 --- comp2Cov11 Covariance term 11 for component 2
(comp2cov11)
317-325 E9.6 --- comp2Cov12 Covariance term 12 for component 2
(comp2cov12)
327-334 E8.6 --- comp2Cov22 Covariance term 22 for component 2
(comp2cov22)
336-343 E8.6 --- comp3Cov11 Covariance term 11 for component 3
(comp3cov11)
345-353 E9.6 --- comp3Cov12 Covariance term 12 for component 3
(comp3cov12)
355-362 E8.6 --- comp3Cov22 Covariance term 22 for component 3
(comp3cov22)
364-372 E9.6 --- comp1Mean1 Mean term 1 for component 1
(comp1mean1)
374-382 E9.6 --- comp1Mean2 Mean term 2 for component 1
(comp1mean2)
384-392 E9.6 --- comp2Mean1 Mean term 1 for component 2
(comp2mean1)
394-402 E9.6 --- comp2Mean2 Mean term 2 for component 2
(comp2mean2)
404-412 E9.6 --- comp3Mean1 Mean term 1 for component 3
(comp3mean1)
414-422 E9.6 --- comp3Mean2 Mean term 2 for component 3
(comp3mean2)
424-431 F8.6 --- comp1Weight Weight for component 1 (comp1_weight)
433-440 F8.6 --- comp2Weight Weight for component 2 (comp2_weight)
442-449 F8.6 --- comp3Weight Weight for component 3 (comp3_weight)
--------------------------------------------------------------------------------
Acknowledgements:
Andres del Pino, apino(at)iaa.es
(End) Patricia Vannier [CDS] 26-Sep-2024