J/AJ/170/287 2min TESS data vetting with Exominer++ (Valizadegan+, 2025)
ExoMiner++: Enhanced transit classification and a new vetting catalog for
2-minute TESS data.
Valizadegan H., Martinho M.J.S., Jenkins J.M., Twicken J.D.,
Caldwell D.A., Maynard P., Wei H., Zhong W., Yates C., Donald S.,
Collins K.A., Latham D., Barkaoui K., Calkins M.L., Carden K., Chazov N.,
Esquerdo G.A., Guillot T., Krushinsky V., Nowak G., Rackham B.V.,
Triaud A., Schwarz R.P., Stephens D., Stockdale C., Watkins C.N.,
Wilkin F.P.
<Astron. J., 170, 287 (2025)>
=2025AJ....170..287V 2025AJ....170..287V
ADC_Keywords: Exoplanets; Photometry, infrared; Optical; Models
Keywords: Exoplanet catalogs ; Exoplanet astronomy ; Exoplanets
Abstract:
We present ExoMiner++, an enhanced deep learning model that builds on
the success of ExoMiner to improve transit signal classification in
2-minute TESS data. ExoMiner++ incorporates additional diagnostic
inputs, including periodogram, flux trend, difference image, unfolded
flux, and spacecraft attitude control data, all of which are crucial
for effectively distinguishing transit signals from more challenging
sources of false positives (FPs). To further enhance performance, we
leverage multisource training by combining high-quality labeled data
from the Kepler space telescope with TESS data. This approach
mitigates the impact of TESS's noisier and more ambiguous labels.
ExoMiner++ achieves high accuracy across various classification and
ranking metrics, significantly narrowing the search space for
follow-up investigations to confirm new planets. To serve the
exoplanet community, we introduce a new TESS catalog containing
ExoMiner++ classifications and confidence scores for each transit
signal. Among the 147,568 unlabeled TCEs, ExoMiner++ identifies 7330
as planet candidates (PCs), with the remainder classified as FPs.
These 7330 PCs correspond to 1868 existing TESS Objects of Interest
(TOIs), 69 Community TESS Objects of Interest (CTOIs), and 50 newly
introduced CTOIs. 1797 out of the 2506 TOIs previously labeled as PCs
in ExoFOP are classified as PCs by ExoMiner++. This reduction in
plausible candidates, combined with the excellent ranking quality of
ExoMiner++, allows the follow-up efforts to be focused on the most
likely candidates, increasing the overall planet yield.
Description:
In the processing of TESS 2min data (IV/39), the workflow comprised
two phases. Initially, data collection spanned years 1 through 4,
encompassing Sectors 1-55 (07/25/18-09/01/22). This yielded
approximately 125,000 SPOC TCEs from 71 transit searches, including 55
single-sector and 16 multi-sector pipeline runs. In the second phase,
data from year 5 Sectors 56-67 (09/01/22-07/29/23) were incorporated,
totaling nearly 208,000 SPOC TCEs across 85 transit searches, with 18
being multi-sector runs. Out of the initial set of 208k TCEs, about
205k were preprocessed successfully. Those that were not preprocessed
had been primarily labeled either as NTPs or UNKs.
We first obtained reliable labels from the publicly available
Exoplanet Follow-up Observing Program (ExoFOP) TESS Object of Interest
(TOI) catalog in 2024 January. This catalog is regularly updated with
new TOIs alerted by the TESS Science Office and then dispositioned by
the TESS Follow-up Observing Program Working Group (TFOPWG) based on a
variety of observations, including but not limited to TESS data. As
the next step for the TCEs without any labels (including PC or APC),
we use the TESS EB catalog from Villanova (Prsa+2022, J/ApJS/258/16).
This is a catalog of about 4500 EBs observed during the first 2yr of
the TESS survey.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table5.dat 542 57162 Scores of different classification model used in
this work on TCEs that are assumed labeled (Model
label can be obtained by score>0.5)
table13.dat 656 147567 Scores and dispositions of TESS+Kepler-Aggregation
model for Unknown (UNK) TCEs
table14.dat 639 2805 Scores and dispositions of TESS+Kepler-Aggregation
model for TOIs that could be matched to TCEs using
ephemeris matching procedure in 4.1.
table16.dat 608 50 List of 50 new CTOIs
--------------------------------------------------------------------------------
See also:
IV/39 : TESS Input Catalog version 8.2 (TIC v8.2) (Paegert+, 2021)
V/133 : Kepler Input Catalog (Kepler Mission Team, 2009)
J/ApJS/217/18 : Potential transit signals in Kepler Q1-Q17 (Seader+, 2015)
J/ApJ/809/77 : Transiting Exoplanet Survey Satellite (TESS) (Sullivan+, 2015)
J/ApJ/822/86 : False positive proba. for Q1-Q17 DR24 KOIs (Morton+, 2016)
J/AJ/158/25 : Automated triage and vetting of TESS candidates (Yu+, 2019)
J/A+A/633/A53 : TESS planet candidates classification (Osborn+, 2020)
J/AJ/161/24 : TRICERATOPS predictions for 384 TOIs (Giacalone+, 2021)
J/ApJS/254/39 : Exoplanet candidates from TESS first 2yr obs (Guerrero+, 2021)
J/ApJS/258/16 : TESS Eclipsing Binary stars. I. Sectors 1-26 (Prsa+, 2022)
J/ApJS/268/4 : Stellar variab. with TESS 2min cadence phot. (Fetherolf+, 2023)
J/AJ/165/95 : Identifying Exoplanets with Deep Learning. V. (Tey+, 2023)
J/AJ/166/28 : Valid. of 69 exoplanets using ExoMiner V1.2 (Valizadegan+, 2023)
Byte-by-byte Description of file: table5.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 19 A19 --- UID Unique ID that includes TIC ID, planet
number, and sector run (G1)
21- 30 I10 --- TIC [358/2051681638] TCE TIC ID
32 I1 --- Num [1/6] TCE planet number
34- 40 F7.2 --- TOI [101.01/6835.01]? TOI number
42 I1 --- Fold [0/9] Cross-validation fold for TCE
44- 57 F14.9 d Per [0.2/1042.06] SPOC TCE Period
59- 70 F12.9 d Dur [6.9e-3/27.16] SPOC TCE Duration
72- 86 F15.9 Rgeo Rad [4.8e-4/19041.54] SPOC TCE Planet Radius
88- 98 E11.9 --- MES [7.1/1.3e11] SPOC TCE Multiple Event
Statistic (MES)
100- 102 A3 --- Lab Original label (1)
104 I1 --- BiLab [0/1] Binary label (2)
106- 116 E11.9 --- Individual [0/1] Score by model trained on TESS
118- 128 E11.9 --- Aggregate [0/1] Aggregate score by model trained on
TESS (3)
130- 140 E11.9 --- IndividualTK [0/1] Score by model trained on
TESS+Kepler (4)
142- 152 E11.9 --- AggregateTK [0/1] Aggregate score by model trained on
TESS+Kepler (3)
154- 281 A128 --- full URL to DV full report
283- 413 A131 --- sum URL to DV summary report
415- 542 A128 --- mini URL to DV mini report
--------------------------------------------------------------------------------
Note (1): Original label as follows:
KP = known planets (1835 occurrences)
CP = confirmed planets (1846 occurrences)
BD = brown dwarf (32 occurrences)
EB = eclipsing binary (12738 occurrences)
FP = false positive (1702 occurrences)
NTP = non-transiting phenomena (39009 occurrences)
Note (2): Binary label as follows:
1 = KP and CP (3681 occurrences)
0 = BD, EB, FP, or NTP (53481 occurrences)
Note (3): Aggregated using strategy in Section 6.3.
Note (4): Multi-source learning idea in Section 6.2.
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table13.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 19 A19 --- UID Unique ID that includes TIC ID, planet
number, and sector run (G1)
21- 30 I10 --- TIC [1886/2055733261] TIC ID
32 I1 --- Num [1/6] TCE planet number
34- 40 F7.2 --- TOI [119.01/6728.01]? TOI number
42- 55 F14.9 d Per [0.19/1754.75] SPOC TCE Period
57- 68 F12.9 d Dur [6.9e-3/22.19] SPOC TCE Duration
70- 84 F15.9 Rgeo Rad [4.9e-5/17113.67] SPOC TCE Planet Radius
86- 96 E11.9 --- MES [7.1/4.2e13] SPOC TCE Multiple Event
Statistic (MES)
98- 100 A3 --- TFOPWG-Jan TFOPWG disposition downloaded in January
2024 (1)
102- 104 A3 --- TFOPWG-Dec TFOPWG disposition downloaded in December
2024 (1)
106- 116 E11.9 --- Score0 [0/1] Score of ExoMiner++ model trained for
fold 0
118- 128 E11.9 --- Score1 [0/1] Score of ExoMiner++ model trained for
fold 1
130- 140 E11.9 --- Score2 [0/1] Score of ExoMiner++ model trained for
fold 2
142- 152 E11.9 --- Score3 [0/1] Score of ExoMiner++ model trained for
fold 3
154- 164 E11.9 --- Score4 [0/1] Score of ExoMiner++ model trained for
fold 4
166- 176 E11.9 --- Score5 [0/1] Score of ExoMiner++ model trained for
fold 5
178- 188 E11.9 --- Score6 [0/1] Score of ExoMiner++ model trained for
fold 6
190- 200 E11.9 --- Score7 [0/1] Score of ExoMiner++ model trained for
fold 7
202- 212 E11.9 --- Score8 [0/1] Score of ExoMiner++ model trained for
fold 8
214- 224 E11.9 --- Score9 [0/1] Score of ExoMiner++ model trained for
fold 9
226- 236 E11.9 --- EScore [0/1] Average score of 10 ExoMiner++ models
238- 248 E11.9 --- EScoreSD [0/0.23] Standard deviation of scores of 10
ExoMiner++ models
250- 260 E11.9 --- EScoreA [0/1] Aggregate score (2)
262- 263 A2 --- ELab ExoMiner++ label (3)
265- 266 A2 --- ELabA ExoMiner++ aggregate label (4)
268- 395 A128 --- full URL to DV full report
397- 527 A131 --- sum URL to DV summary report
529- 656 A128 --- mini URL to DV mini report
--------------------------------------------------------------------------------
Note (1): TFOPWG disposition as follows:
PC = planet candidate (7990 occurrences in Jan, 7685 in Dec)
CP = confirmed planets (0 occurrence in Jan, 151 in Dec)
FA = false alarm (77 occurrences in Jan, 83 in Dec)
APC = ambiguous planet candidate (1050 occurrences in Jan, 1056 in Dec)
FP = false positive (0 occurrence in Jan, 142 in Dec)
Note (2): Computed on top of tess-individual mean score.
Note (3): Exominer++ label as follows:
PC = if ExoMiner++ score>0.5 (7054 occurrences)
FP = if ExoMiner++ score≤0.5 (140513 occurrences)
Note (4): Exominer aggregate label as follows:
PC = if ExoMiner++ aggregate score>0.5 (140237 occurrences)
FP = if ExoMiner++ aggregate score≤0.5 (7330 occurrences)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table14.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 19 A19 --- UID Unique ID that includes TIC ID, planet
number, and sector run (G1)
21- 30 I10 --- TIC [1018843/1980725947] TIC ID
32 I1 --- Num [1/5] TCE planet number
34- 40 F7.2 --- TOI [119.01/6728.01] TOI ID
42- 55 F14.9 d Per [0.23/1467.99] SPOC TCE Period
57- 67 F11.9 d Dur [0.01/0.81] SPOC TCE Duration
69- 80 F12.9 Rgeo Rad [0.55/54.59] SPOC TCE Planet Radius
82- 94 F13.9 --- MES [7.12/466.28] SPOC TCE Multiple Event
Statistic (MES)
96- 98 A3 --- TFOPWG-Jan TFOPWG disposition downloaded in January
2024 (1)
100- 102 A3 --- TFOPWG-Dec TFOPWG disposition downloaded in December
2024 (1)
104- 114 F11.9 --- Score0 [1.78e-6/1] Score of ExoMiner++ model
trained for fold 0
116- 126 F11.9 --- Score1 [5.36e-7/1] Score of ExoMiner++ model
trained for fold 1
128- 138 F11.9 --- Score2 [9.77e-7/1] Score of ExoMiner++ model
trained for fold 2
140- 150 F11.9 --- Score3 [5.12e-6/1] Score of ExoMiner++ model
trained for fold 3
152- 162 F11.9 --- Score4 [4.44e-6/1] Score of ExoMiner++ model
trained for fold 4
164- 174 F11.9 --- Score5 [1.11e-6/1] Score of ExoMiner++ model
trained for fold 5
176- 186 F11.9 --- Score6 [1e-5/1] Score of ExoMiner++ model trained
for fold 6
188- 198 F11.9 --- Score7 [9.43e-7/1] Score of ExoMiner++ model
trained for fold 7
200- 210 F11.9 --- Score8 [9.43e-6/1] Score of ExoMiner++ model
trained for fold 8
212- 222 F11.9 --- Score9 [2.57e-5/1] Score of ExoMiner++ model
trained for fold 9
224- 234 F11.9 --- EScore [1.36e-5/1] Average score of 10 ExoMiner++
models
236- 246 F11.9 --- EScoreSD [1.57e-5/0.22] Standard deviation of scores
of 10 ExoMiner++ models
248- 249 A2 --- ELab ExoMiner++ label (2)
251- 378 A128 --- full URL to DV full report
380- 510 A131 --- sum URL to DV summary report
512- 639 A128 --- mini URL to DV mini report
--------------------------------------------------------------------------------
Note (1): TFOPWG disposition as follows:
PC = planet candidate (2506 occurrences in Jan, 2423 in Dec)
CP = confirmed planets (0 occurrence in Jan, 32 in Dec)
FA = false alarm (43 occurrences in Jan, 45 in Dec)
APC = ambiguous planet candidate (254 occurrences in Jan, 263 in Dec)
FP = false positive (0 occurrence in Jan, 40 in Dec)
Note (2): Exominer++ label as follows:
PC = if ExoMiner++ score>0.5 (1868 occurrences)
FP = if ExoMiner++ score≤0.5 (937 occurrences)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table16.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 19 A19 --- UID Unique ID that includes TIC ID, planet
number, and sector run (G1)
21- 30 I10 --- TIC [5892614/1715469667] TIC ID
32 I1 --- Num [1/3] Planet number
34- 45 F12.9 d Per [0.23/49.63] SPOC TCE Period
47- 57 F11.9 d Dur [0.02/0.58] SPOC TCE Duration
59- 70 F12.9 Rgeo Rad [0.77/23.02] SPOC TCE Planet Radius
72- 84 F13.9 --- MES [7.25/229.31] SPOC TCE Multiple Event
Statistic (MES)
86- 95 F10.8 --- Score0 [0.53/1] Score of ExoMiner++ model trained
for fold 0
97- 106 F10.8 --- Score1 [0.53/1] Score of ExoMiner++ model trained
for fold 1
108- 117 F10.8 --- Score2 [0.55/1] Score of ExoMiner++ model trained
for fold 2
119- 128 F10.8 --- Score3 [0.52/0.99] Score of ExoMiner++ model trained
for fold 3
130- 139 F10.8 --- Score4 [0.56/1] Score of ExoMiner++ model trained
for fold 4
141- 150 F10.8 --- Score5 [0.52/1] Score of ExoMiner++ model trained
for fold 5
152- 161 F10.8 --- Score6 [0.52/1] Score of ExoMiner++ model trained
for fold 6
163- 172 F10.8 --- Score7 [0.5/1] Score of ExoMiner++ model trained for
fold 7
174- 183 F10.8 --- Score8 [0.55/1] Score of ExoMiner++ model trained
for fold 8
185- 194 F10.8 --- Score9 [0.52/1] Score of ExoMiner++ model trained
for fold 9
196- 206 F11.9 --- EScore [0.58/1] Average score of 10 ExoMiner++
models
208- 218 F11.9 --- EScoreSD [2.9e-3/0.12] Standard deviation of scores of
10 ExoMiner++ models
220- 347 A128 --- full URL to DV full report
349- 479 A131 --- sum URL to DV summary report
481- 608 A128 --- mini URL to DV mini report
--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
Global notes:
Note (G1): To refer to a TESS SPOC TCE, we use the naming form of x-y-Sz where
x is the target TIC ID, y is the TCE planet number set by the SPOC pipeline,
and z is the sector run (if TCE comes from a multi-sector run, then Sz1-z2,
where z1 and z2 are the start and end sectors of the sector run, respectively)
where it was detected.
--------------------------------------------------------------------------------
History:
From electronic version of the journal
License: CC-BY-4.0
(End) Prepared by [ASS], Robin Leichtnam [CDS] 17-Aug-2026