J/AJ/158/25 Automated triage and vetting of TESS candidates (Yu+, 2019)
Identifying exoplanets with deep learning.
III. Automated triage and vetting of TESS candidates.
Yu L., Vanderburg A., Huang C., Shallue C.J., Crossfield I.J.M.,
Gaudi B.S., Daylan T., Dattilo A., Armstrong D.J., Ricker G.R.,
Vanderspek R.K., Latham D.W., Seager S., Dittmann J., Doty J.P.,
Glidden A., Quinn S.N.
<Astron. J., 158, 25-25 (2019)>
=2019AJ....158...25Y 2019AJ....158...25Y (SIMBAD/NED BibCode)
ADC_Keywords: Stars, double and multiple ; Exoplanets ; Magnitudes ; Models
Keywords: methods: data analysis - planets and satellites: detection -
techniques: photometric
Abstract:
NASA's Transiting Exoplanet Survey Satellite (TESS) presents us with
an unprecedented volume of space-based photometric observations that
must be analyzed in an efficient and unbiased manner. With at least
∼1000000 new light curves generated every month from full-frame images
alone, automated planet candidate identification has become an attractive
alternative to human vetting. Here we present a deep learning model
capable of performing triage and vetting on TESS candidates. Our model
is modified from an existing neural network designed to automatically
classify Kepler candidates, and is the first neural network to be trained
and tested on real TESS data. In triage mode, our model can distinguish
transit-like signals (planet candidates and eclipsing binaries) from
stellar variability and instrumental noise with an average precision
(the weighted mean of precisions over all classification thresholds)
of 97.0% and an accuracy of 97.4%. In vetting mode, the model is trained
to identify only planet candidates with the help of newly added scientific
domain knowledge, and achieves an average precision of 69.3% and an
accuracy of 97.8%. We apply our model on new data from Sector 6, and
present 288 new signals that received the highest scores in triage and
vetting and were also identified as planet candidates by human vetters.
We also provide a homogeneously classified set of TESS candidates suitable
for future training.
Description:
Here we present the first Convolutional neural network (CNN) trained and
tested on real TESS data. Our model takes as inputs human-labeled light
curves produced by the MIT Quick Look Pipeline (QLP; C. Huang et al.
2019, in preparation), and can be trained to perform either triage or
vetting on TESS candidates. Like Shallue & Vanderburg (2018AJ....155...94S 2018AJ....155...94S),
we work with possible planet signals, which are called "threshold-crossing
events" or TCEs. These are periodic dimming events potentially consistent
with signals produced by transiting planets, and are typically identified
by an algorithm designed to find such signals. In this study, we adopt
the MIT QLP for light-curve production and transit searches.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table2.dat 64 288 New threshold-crossing events (TCEs) from
Sector 6 with the highest likelihood of being
planet candidates and manually assigned planet
candidates (PC) labels
--------------------------------------------------------------------------------
See also:
IV/38 : TESS Input Catalog - v8.0 (TIC-8) (Stassun+, 2019)
J/ApJ/809/77 : Transiting Exoplanet Survey Satellite (TESS) (Sullivan+, 2015)
J/AJ/156/102 : TESS Input Catalog and Candidate Target List (Stassun+, 2018)
J/ApJS/239/2 : Simulated exoplanets from TESS list of targets (Barclay+, 2018)
Byte-by-byte Description of file: table2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 11 I11 --- TIC [4616346/10001673159] TESS Input Catalog identifier
13- 18 F6.3 mag Tmag [6.781/11.995] TESS-band magnitude
20- 28 F9.6 d Per [0.259746/13.9862] Period
30- 40 F11.6 d T0 [1325.7/1481.82] Epoch of transit (BJDTDB-2457000)
42- 46 F5.2 h Dur [0.74/11.43] Transit duration
48- 52 I5 ppm Depth [190/38350] Transit depth
54- 58 F5.3 --- Tri [0.164/0.998] Triage prediction
60- 64 F5.3 --- Vet [0.1/0.935] Vetting prediction
--------------------------------------------------------------------------------
History:
From electronic version of the journal
References:
Shallue et al. Paper I. 2018AJ....155...94S 2018AJ....155...94S
Dattilo et al. Paper II. 2019AJ....157..169D 2019AJ....157..169D
(End) Tiphaine Pouvreau [CDS] 29-Aug-2019