Identifying Exoplanets with Deep Learning. V. Improved Light-curve Classification for TESS Full-frame Image Observations
Article
Tey, Evan, Dan, Moldovan, Kunimoto, Michelle, Huang, Chelsea X., Shporer, Avi, Daylan, Daylan, Muthukrishna, Daniel, Vanderburg, Andrew, Dattilo, Anne, Ricker, George R. and Seager, S.. 2023. "Identifying Exoplanets with Deep Learning. V. Improved Light-curve Classification for TESS Full-frame Image Observations." The Astronomical Journal. 165, p. 95. https://doi.org/10.3847/1538-3881/acad85
Article Title | Identifying Exoplanets with Deep Learning. V. Improved Light-curve Classification for TESS Full-frame Image Observations |
---|---|
ERA Journal ID | 1048 |
Article Category | Article |
Authors | Tey, Evan, Dan, Moldovan, Kunimoto, Michelle, Huang, Chelsea X., Shporer, Avi, Daylan, Daylan, Muthukrishna, Daniel, Vanderburg, Andrew, Dattilo, Anne, Ricker, George R. and Seager, S. |
Journal Title | The Astronomical Journal |
Journal Citation | 165, p. 95 |
Number of Pages | 95 |
Year | 2023 |
Publisher | IOP Publishing |
Place of Publication | United States |
ISSN | 0004-6256 |
1538-3881 | |
Digital Object Identifier (DOI) | https://doi.org/10.3847/1538-3881/acad85 |
Web Address (URL) | https://iopscience.iop.org/article/10.3847/1538-3881/acad85 |
Abstract | The TESS mission produces a large amount of time series data, only a small fraction of which contain detectable exoplanetary transit signals. Deep-learning techniques such as neural networks have proved effective at differentiating promising astrophysical eclipsing candidates from other phenomena such as stellar variability and systematic instrumental effects in an efficient, unbiased, and sustainable manner. This paper presents a high-quality data set containing light curves from the Primary Mission and 1st Extended Mission full-frame images and periodic signals detected via box least-squares. The data set was curated using a thorough manual review process then used to train a neural network called Astronet-Triage-v2. On our test set, for transiting/eclipsing events, we achieve a 99.6% recall (true positives over all data with positive labels) at a precision of 75.7% (true positives over all predicted positives). Since 90% of our training data is from the Primary Mission, we also test our ability to generalize on held-out 1st Extended Mission data. Here, we find an area under the precision-recall curve of 0.965, a 4% improvement over Astronet-Triage. On the TESS object of interest (TOI) Catalog through 2022 April, a shortlist of planets and planet candidates, Astronet-Triage-v2 is able to recover 3577 out of 4140 TOIs, while Astronet-Triage only recovers 3349 targets at an equal level of precision. In other words, upgrading to Astronet-Triage-v2 helps save at least 200 planet candidates from being lost. The new model is currently used for planet candidate triage in the Quick-Look Pipeline. |
Keywords | Neural networks; Transit photometry; Exoplanet detection methods; |
ANZSRC Field of Research 2020 | 510109. Stellar astronomy and planetary systems |
Byline Affiliations | Massachusetts Institute of Technology, United States |
Google, United States | |
Centre for Astrophysics | |
Princeton University, United States | |
University of California Berkeley, United States |
Permalink -
https://research.usq.edu.au/item/yy945/identifying-exoplanets-with-deep-learning-v-improved-light-curve-classification-for-tess-full-frame-image-observations
Download files
49
total views29
total downloads9
views this month1
downloads this month