J/MNRAS/483/5077 Machine learning to identify symbiotic stars (Akras+, 2019)
A machine learning approach for identification and classification of symbiotic
stars using 2MASS and WISE.
Akras S., Leal-Ferreira M.L., Guzman-Ramirez L., Ramos-Larios G.
<Mon. Not. R. Astron. Soc., 483, 5077-5104 (2019)>
=2019MNRAS.483.5077A 2019MNRAS.483.5077A (SIMBAD/NED BibCode)
ADC_Keywords: Stars, double and multiple ; Stars, fundamental ;
Photometry, classification ; Photometry, H-alpha
Keywords: methods: data analysis - methods: statistical - general: catalogues -
stars: binaries: symbiotic - stars: fundamental parameters
Abstract:
In this second paper in a series of papers based on the
most-up-to-date catalogue of symbiotic stars (SySts), we present a new
approach for identifying and distinguishing SySts from other Hα
emitters in photometric surveys using machine learning algorithms such
as classification tree, linear discriminant analysis, and K-nearest
neighbour. The motivation behind this work is to seek for possible
colour indices in the regime of near- and mid-infrared covered by the
2MASS and WISE surveys. A number of diagnostic colour-colour diagrams
are generated for all the known Galactic SySts and several classes of
stellar objects that mimic SySts such as planetary nebulae, post-AGB,
Mira, single K and M giants, cataclysmic variables, Be, AeBe, YSO,
weak and classical T Tauri stars, and Wolf-Rayet. The classification
tree algorithm unveils that primarily J-H, W1-W4, and Ks-W3, and
secondarily, H-W2, W1-W2, and W3-W4 are ideal colour indices to
identify SySts. Linear discriminant analysis method is also applied to
determine the linear combination of 2MASS and AllWISE magnitudes that
better distinguish SySts. The probability of a source being an SySt is
determined using the K-nearest neighbour method on the LDA components.
By applying our classification tree model to the list of candidate
SySts (Paper I, Akras et al. 2019MNRAS.483.5077A 2019MNRAS.483.5077A), the IPHAS list of
candidate SySts, and the DR2 VPHAS+ catalogue, we find 125 (72 new
candidates) sources that pass our criteria while we also recover 90
per cent of the known Galactic SySts.
Description:
We carried out and presented a machine learning approach to find new
SySts in publicly available Hα photometric catalogues using
Hα-excess, 2MASS (Skrutskie et al. 2006AJ....131.1163S 2006AJ....131.1163S, Cat.
VII/233), and WISE (Wright et al. 2010AJ....140.1868W 2010AJ....140.1868W, Cat. II/311)
photometric data. First, we explored a number of different
combinations of colour indices that can provide a good separation of
SySts from other classes of objects that mimic SySts such as PNe,
post-AGB stars, CVs, WR stars, WTT and ClTT stars, single K and M
giants, and Be stars. We shown that the widely used J-H versus H-Ks is
not an adequate diagnostic colour-colour diagram (DCCD) for
identifying SySts. S-type SySts, Mira, YSO, and WTT stars occupy the
same regions making very hard to distinguish them. The W3-W4 versus
Ks-W3 and J-H versus W1-W4 were found to be better DCCDs.
The list of 125 sources found in the list of the candidate SySts
(Paper I, Akras et al. 2019ApJS..240...21A 2019ApJS..240...21A, Cat. J/ApJS/240/21), the
IPHAS list of candidate SySts (Corradi et al. 2008A&A...480..409C 2008A&A...480..409C,
Cat. J/A+A/480/409), and the DR2 VPHAS + catalogue (Drew et al.
2014MNRAS.440.2036D 2014MNRAS.440.2036D, Cat. J/MNRAS/440/2036) are presented in tables C1
and C2.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
tablec1.dat 117 125 A further classification of the candidates
based on the classification tree
tablec2.dat 166 125 A further classification of the candidates
based on the LDA/KNN
--------------------------------------------------------------------------------
See also:
VII/233 : The 2MASS Extended sources (IPAC/UMass, 2003-2006)
II/311 : WISE All-Sky Data Release (Cutri+ 2012)
II/341 : VPHAS+ DR2 survey (Drew+, 2016)
J/A+A/480/409 : IPHAS symbiotic stars candidates (Corradi+, 2008)
J/MNRAS/440/2036 : VPHAS+ survey synthetic colours (Drew+, 2014)
J/ApJS/240/21 : Symbiotic stars with 2MASS, WISE + Gaia data (Akras+, 2019)
Byte-by-byte Description of file: tablec1.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 7 A7 --- Class Symbiotic star classification (G1)
9- 34 A26 --- Name Candidate SySt name
36- 39 A4 --- SPB SySts/PNe/Be classification (G2)
41- 44 A4 --- SCM SySts/CV/Mira classification (G3)
46- 49 A4 --- SCY SySts/CV/YSO classification (G4)
51- 54 A4 --- SWpA SySts/WR/post-AGB classification (G5)
56- 61 A6 --- SKM SySts/K-giants/M-giants classification (G6)
63- 70 A8 --- SWC SySts/WTT/ClTT classification (G7)
72- 75 A4 --- SBA SySts/Be/AeBe classification (G8)
77- 98 A22 --- OName Other name found in literature for the SySt
candidate
100-113 A14 --- Type Comments on the SySt candidate (G9)
115-117 A3 --- r_Type References for Type (G10)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: tablec2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 7 A7 --- Class Symbiotic star classification (G1)
9- 34 A26 --- Name Candidate SySt name
36- 46 A11 --- SPB SySts/PNe/Be classification (G2)
48- 58 A11 --- SCM SySts/CV/Mira classification (G3)
60- 70 A11 --- SCY SySts/CV/YSO classification (G4)
72- 82 A11 --- SWpA SySts/WR/post-AGB classification (G5)
84- 96 A13 --- SKM SySts/K-giants/M-giants classification (G6)
98-112 A15 --- SWC SySts/WTT/ClTT classification (G7)
114-124 A11 --- SBA SySts/Be/AeBe classification (G8)
126-147 A22 --- OName Other name found in literature for the SySt
candidate
149-162 A14 --- Type Comments on the SySt candidate (G9)
164-166 A3 --- r_Type References for Type (G10)
--------------------------------------------------------------------------------
Global Notes:
Note (G1): Class as follows:
stellar = S-type SySt
dusty = D-type SySt
Note (G2): The first training sample includes SySts, PNe, and Be, two of the
most common mimic of SySts due to the emission of several common
lines. The first W1-W4 colour index criterion discriminates SySts
and Be stars from PNe. Almost all PNe (166 sources or 88 per cent)
satisfy the criterion W1-W4>7.285 while they suffer by a 3 per cent
contamination from Be and SySts. 98 per cent of Be stars show
J-H<0.541. On the other hand, 96 per cent of the known Galactic SySts
show J-H>0.541 and W3-W4<2.56 suffering of only 3 per cent
contamination.
Note (G3): Due to the upper limit magnitudes of CVs in W3 and W4, we did not use
the colours W2-W3, W3-W4, W1-W4, and Ks-W3). SySts are separated into
two groups depending on the W1-W2 colour index. SySts with
W1-W2<0.151 are classified as S-type and they are systematically
redder in the J-H colour than CVs. On the other side, the dusty SySts
exhibit W1-W2>0.151 but they are in the minority compared with Mira
stars and CVs which show clearly different J-H colours (CVs:
J-H<0.684, Mira: J-H>0.684).
Note (G4): This training sample contains SySts, CV, and YSO. In this case, CVs
are easily separated from SySts and YSO based on the criterion
J-H<0.663 and the upper limit magnitudes in W3 and W4 do not affect
our model. SySts and YSO exhibit both J-H>0.663 and they are
distinguished based on the Ks-W3 colour. SySts with Ks-W3<1.344
correspond to S-type whereas those with W3-W4>1.344 correspond to the
dusty SySts and are mixed with YSO.
Note (G5): SySts, post-AGB, and WR stars can also be separated very well using
the J-H and W1-W4 colour indices. The vast majority of post-AGB stars
show W1-W4>4.735 and they are contaminated by only few WR and SySts.
SySts (92 per cent) are found to exhibit W1-W4<4.735 and J-H>0.774
while WR stars have J-H<0.774.
Note (G6): Given that S-type SySts have a M or K giant companion, it is coherent
to explore the colour indices that discriminate SySts from single M/K
giants. Almost all SySts (95 per cent) have H-W2>0.206 and Ks-W3>0.27
with a very small contamination mostly from M giants and few K giants
(3 per cent), which should be further investigated. Approximately 50
per cent of M giants show the same H-W2 colour with SySts but are
bluer in the Ks-W3 colour index (<0.27). Regarding the K giants, the
majority of them shows H-W2 colour index <0.206 and they are
separated from M giants based on the W2-W3 colour index.
Note (G7): Almost all SySts have W3-W4<1.483 and J-H>0.78 and they suffer of
only 5 per cent contamination. Half of WTT stars are found to be
bluer in the J-H colour index (<0.78) compared to SySts. The
remaining of WTT have W3-W4>1.483 and they are mixed with ClTT and
few SySts. The W1-W2 colour index separates further the WTT and the
ClTT stars with the former being bluer and the latter redder.
Note (G8): The W1-W4 colour index is the first criterion that strongly
discriminated SySts and Be stars from AeBe stars. SySts and Be are
found to be bluer in the W1-W4 colour (<3.949) compared to the AeBe
stars (>3.949). SySts and Be stars are further separated based on the
J-H colour index (Be<0.63, SySts>0.63). The contamination of these
two groups is small of the order of 5.9 and 1.7 per cent,
respectively. On the other hand, AeBe stars exhibit W1-W2>0.03 with
a very small contamination of SySts and Be.
Note (G9): The classification of some sources as emission line stars (ELS),
semiregular pulsating star (SR-PS), asymptotic giant branch stars
(AGB), Wolf-Rayet stars (WR), planetary nebula (PN) or
known/candidate symbiotic stars (SySt) is based on the SIMBAD
catalogue, Kohoutek & Wehmeyer (2003AN....324..437K 2003AN....324..437K,
Cat. J/AN/324/437) or Paper I (Akras et al. 2019ApJS..240...21A 2019ApJS..240...21A,
Cat. J/ApJS/240/21).
Note (G10): References as follows:
1 = Rodriguez-Flores et al. (2014A&A...567A..49R 2014A&A...567A..49R, Cat. J/A+A/567/A49)
2 = Corradi et al. (2010A&A...509A..41C 2010A&A...509A..41C)
3 = Viironen et al. (2009A&A...504..291V 2009A&A...504..291V, Cat. J/A+A/504/291)
4 = Krause et al. (2003A&A...398.1007K 2003A&A...398.1007K)
--------------------------------------------------------------------------------
History:
From electronic version of the journal
References:
Akras et al., Paper I 2019ApJS..240...21A 2019ApJS..240...21A, Cat. J/ApJS/240/21
(End) Ana Fiallos [CDS] 05-Aug-2022