J/A+A/633/A53 TESS planet candidates classification (Osborn+, 2020)
Rapid classification of TESS planet candidates with convolutional neural
networks.
Osborn H.P., Ansdell M., Ioannou Y., Sasdelli M., Angerhausen D.,
Caldwell D.A., Jenkins J.M., Raissi C., Smith J.C.
<Astron. Astrophys. 633, A53 (2020)>
=2020A&A...633A..53O 2020A&A...633A..53O (SIMBAD/NED BibCode)
ADC_Keywords: Stars, double and multiple ; Planets
Keywords: planets and satellites: detection - methods: analytical
Abstract:
Accurately and rapidly classifying exoplanet candidates from transit
surveys is a goal of growing importance as the data rates from
space-based survey missions increase. This is especially true for
NASA's TESS mission which generates thousands of new candidates each
month. Here we created the first deep learning model capable of
classifying TESS planet candidates.
We adapted the neural network model of Ansdell et al (2018) to TESS
data. We then trained and tested this updated model on 4 sectors of
high-fidelity, pixel-level simulations data created using the Lilith
simulator & processed using the full TESS pipeline. With the caveat
that direct transfer of the model to real data will not perform as
accurately, we also applied this model to four sectors of TESS
candidates.
We find our model performs very well on our simulated data, with 97%
average precision and 92% accuracy on planets in the 2-class model.
This accuracy is also boosted by another ∼4% if planets found at the
wrong periods are included. We also performed 3- and 4-class
classification of planets, blended & target eclipsing binaries, and
non-astrophysical false positives, which have slightly lower average
precision and planet accuracies, but are useful for follow-up
decisions. When applied to real TESS data, 61% of Threshold Crossing
Events (TCEs) coincident with currently published TOIs are recovered
as planets, 4% more are suggested to be Eclipsing Binaries, and we
propose a further 200 TCEs as planet candidates.
Description:
Predictions of Threshold Crossing Events (TCEs) identified in the
first 5 sectors of NASA's Transiting Exoplanet Survey Satellite (TESS)
using our Neural Network models which have been trained on four
simulated sectors ofTESS lightcurves. The models include binary
(2-class) planet/non-planet classification, a 3-class model including
Planets, Eclipsing Binaries and dips of Non-astrophysical (or unknown)
origin, and a 4-class model with EBs split into target EBs and
blended/background EBs.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
tceclass.dat 952 7562 Class probability estimates for each
Threshold Crossing Event (TCE)
(Tables A1 and A2)
--------------------------------------------------------------------------------
See also:
J/AJ/156/102 : TESS Input Catalog and Candidate Target List (Stassun+, 2019)
Byte-by-byte Description of file: tceclass.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 A18 --- Name Unique Threshold Crossing Event (TCE)
identifier, in the form TIDndetsector
(IDTCE)
20- 41 E22.17 --- P2med Planet class probability (2-class model,
median average over ensemble of models)
(binary_med)
43- 64 E22.17 --- P2av Planet class probability (2-class model, mean
average over ensemble of models) (binary_av)
66- 87 E22.17 --- P3UNKmed Unknown class probability (3-class model,
median average over ensemble of models)
(3classUNKmed)
89-110 E22.17 --- P3UNKav Unknown class probability (3-class model, mean
average over ensemble of models)
(3classUNKav)
112-133 E22.17 --- P3PLmed Planet class probability (3-class model,
median average over ensemble of models)
(3classPLmed)
135-156 E22.17 --- P3PLav Planet class probability (3-class model, mean
average over ensemble of models)
(3classPLav)
158-179 E22.17 --- P3EBmed EB class probability (3-class model, median
average over ensemble of models)
(3classEBmed)
181-202 E22.17 --- P3EBav EB class probability (3-class model, mean
average over ensemble of models)
(3classEBav)
204-225 E22.17 --- P4UNKme Unknown class probability (4-class model,
median average over ensemble of models)
(4classUNKmed)
227-248 E22.17 --- P4UNKav Unknown class probability (4-class model, mean
average over ensemble of models)
(4classUNKav)
250-271 E22.17 --- P4PLmed Planet class probability (4-class model,
median average over ensemble of models)
(4classPLmed)
273-294 E22.17 --- P4PLav Planet class probability (4-class model, mean
average over ensemble of models)
(4classPLav)
296-317 E22.17 --- P4EBmed EB class probability (4-class model, median
average over ensemble of models)
(4classEBmed)
319-340 E22.17 --- P4EBav EB class probability (4-class model, mean
average over ensemble of models)
(4classEBav)
342-363 E22.17 --- P4BEBme BEB class probability (4-class model, median
average over ensemble of models)
(4classBEBmed)
365-386 E22.17 --- P4BEBav BEB class probability (4-class model, mean
average over ensemble of models)
(4classBEBav)
388-392 A5 --- Nans [False/True ] Is the input data affected by
Nans? (nans)
394-415 E22.17 --- Pplavall Planet class probability (mean average over
all 2-,3- & 4-class models) (plavall)
417-438 E22.17 --- Pplmedall Planet class probability (median average over
all 2-,3- & 4-class models) (plmedall)
440-443 I4 --- Rank Ranked position in plavall column (rank)
445-469 F25.16 --- MaxMES Max Multiple Event Statistic (total SNR)
(MAXMES)
471-495 F25.16 --- MaxSES Max Single Event Statistic (individual transit
SNRs) (MAXSES)
497-502 I6 ppm Tdepth TCE Depth (TDEPTH)
504-524 F21.17 h Tdur TCE Duration (TDUR)
526-543 F18.13 d Tepoch Epoch of first transit (BJD-2454833.0)
(TEPOCH)
545-564 F20.17 d Tperiod TCE Period (TPERIOD)
566-587 E22.17 --- TSNR ? TCE in-transit SNR (TSNR)
589-590 I2 --- ndet [0/11] Index of Threshold Crossing Event (TCE)
for this TID (n_det)
592-593 I2 --- Sector TESS sector (1-4, 99=multi-sector) (sector)
595-599 A5 --- Sectors [1 2 3 4] TESS sectors searched (sectors)
601-604 F4.2 --- is-TOI [0.0/1] TCE matches identified TOI
(1 = good match) (is_toi)
606-611 F6.2 --- TOI [101.01/419.01]?=0 Tess Object of Interest
number (toi)
613-615 A3 --- TOIdisp [0.0 O PC] TOI disposition (toi_disp)
617-625 I9 --- TIC Tess ID (from Tess input catalogue) (TID)
627-637 F11.1 --- ID ID (from Tess input catalogue), TIC.0 (ID)
639-660 F22.19 [-] [M/H] ? Target star Metallicity (MH)
662-669 F8.2 K Teff ? Target star effective surface temperature
(Teff)
671-689 F19.16 mag Tmag Target star Tess magnitude (Tmag)
691-712 E22.15 --- contratio ? Target star contamination ratio (contratio)
714-735 F22.15 pc Dist ? Target star distance (d)
737-756 F20.16 deg DEdeg Target star declination (J2000) (dec)
758-776 F19.15 deg GLAT Target star galactic latitude (gallat)
778-797 F20.16 deg GLON Target star galactic longitude (gallong)
799-816 F18.16 [cm/s2] logg ? Target star log surface gravity (logg)
818-836 F19.17 Msun Mass ? Target star mass in solar masses (mass)
838-841 A4 --- objType [STAR] Target star object type (objType)
843-864 F22.17 mas/yr pmDE ? Target star proper motion in declination
(pmDEC)
866-888 F23.18 mas/yr pmRA* ? Target star proper motion in RA,
pmRA*cosDE (pmRA)
890-909 F20.16 deg RAdeg Target star right ascension (J2000) (ra)
911-930 F20.17 Rsun rad ? Target star radii in solar radius (rad)
932-952 E21.17 Sun rho ? Target star density scaled to solar density
(rho)
--------------------------------------------------------------------------------
Acknowledgements:
Hugh Osborn, hugh.osborn(at)lam.fr
(End) Patricia Vannier [CDS] 09-Apr-2019