Abstract
The Kepler and Transiting Exoplanet Survey Satellite (TESS) missions have generated over 100,000 potential transit signals that must be processed in order to create a catalog of planet candidates. During the past few years, there has been a growing interest in using machine learning to analyze these data in search of new exoplanets. Different from the existing machine learning works, ExoMiner, the proposed deep learning classifier in this work, mimics how domain experts examine diagnostic tests to vet a transit signal. ExoMiner is a highly accurate, explainable, and robust classifier that (1) allows us to validate 301 new exoplanets from the MAST Kepler Archive and (2) is general enough to be applied across missions such as the ongoing TESS mission. We perform an extensive experimental study to verify that ExoMiner is more reliable and accurate than the existing transit signal classifiers in terms of different classification and ranking metrics. For example, for a fixed precision value of 99%, ExoMiner retrieves 93.6% of all exoplanets in the test set (i.e., recall = 0.936), while this rate is 76.3% for the best existing classifier. Furthermore, the modular design of ExoMiner favors its explainability. We introduce a simple explainability framework that provides experts with feedback on why ExoMiner classifies a transit signal into a specific class label (e.g., planet candidate or not planet candidate).
| Original language | English |
|---|---|
| Article number | 120 |
| Journal | Astrophysical Journal |
| Volume | 926 |
| Issue number | 2 |
| DOIs | |
| State | Published - 1 Feb 2022 |
Bibliographical note
Funding Information:H.V., M.M., L.W., and N.W. are suported through NASA NAMS contract NNA16BD14C. J.S., D.C., and J.T. are supported through NASA Cooperative Agreement 80NSSC21M0079. We would like to thank multiple people who directly or indirectly contributed to this work. We are very grateful to Porsche M. Parker and Haley E. Feck from USRA and Krisstina Wilmoth from NASA ARC, who have helped us tirelessly to recruit interns through the NASA ARC I intern program and NASA ARC Office of STEM Engagement. Without these amazing interns, this work would not have been possible. We are grateful to Patrick Maynard, who has recently joined our team as an intern and generated new insights into TESS data. We also appreciate David J. Armstrong for providing us with the detailed results of their ML models (Armstrong et al. ) and Megan Ansdell, who answered our questions regarding the ExoNet code (Ansdell et al. ). Our discussion with David J. Armstrong on the caveats of using ML for this problem helped us tremendously in improving this work. This paper includes data collected by the Kepler and TESS missions and obtained from the MAST data archive at the Space Telescope Science Institute (STScI). Funding for the Kepler mission is provided by the NASA Science Mission Directorate. Funding for the TESS mission is provided by the NASA Explorer Program. STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 526555. We acknowledge the use of public TESS data from pipelines at the TESS Science Office and at the TESS Science Processing Operations Center. Resources supporting this work were provided by the NASA High-End Computing (HEC) Program through the NASA Advanced Supercomputing (NAS) Division at Ames Research Center for the production of the SOC Kepler and the SPOC TESS data products. 2
Publisher Copyright:
© 2022. The Author(s). Published by the American Astronomical Society.