NASA Logo, National Aeronautics and Space Administration

+NASA Home

+Ames Home

TESS Team Develops Automated Discovery of New Planets Capability Using Deep Learning
Intelligent Systems Division Banner

TESS Team Develops Automated Discovery of New Planets Capability Using Deep Learning

The Terrestrial Exoplanet Survey Satellite (TESS) Science Processing Operations Center (SPOC) team has recently developed ExoMiner, a new deep neural network that has now been used to validate 301 new exoplanets. ExoMiner utilizes the unique elements of the Kepler Science Operations Center (SOC)/TESS SPOC data validation summary report in its original format to classify transit signals. This is unlike existing machine classifiers that either do not use a comprehensive list of diagnostic tests required for correct classification or use a simplified version of these tests in the form of a few scalar values.

This work was published in the Astrophysical Journal in February 2022 and has been recognized by the Universities Space Research Association (USRA), NASA Ames, NASA’s Jet Propulsion Laboratory (JPL), the Search for ExtraTerrestrial Intelligence (SETI) Institute, and featured on the NASA Exoplanet website in addition to being covered by many news media outlets, such as pbs.org, yahoo.com, newsweek.com, newyorkweeklytimes.com, universetoday.com, spaceref.com, voanews.com, thehill.com, weather.com, earthsky.org, space.com, bussinessinsider.com, and many others.

BACKGROUND: Kepler and TESS mission work is based on transit photometry, where a target star is monitored for periodic dimming (called transiting events) of its brightness. This dimming could be caused by an exoplanet or some other sources (false positives). The manual vetting of these transiting events is very time consuming, which led to development of a machine-learning vetting system in order to classify transit signals into planet candidates and false positives. ExoMiner is the result of several years of research in automating this process.

NASA PROGRAM FUNDING: Miguel Martinho, Hamed Valizadegan, Nikash Walia, and Laurent Wilkens are supported through NASA Academic Mission Services (NAMS) Contract NNA16BD14C. Douglas Caldwell, Jeffrey Smith, and Joseph Twicken are supported through NASA Cooperative Agreement 80NSSC21M0079.

TEAM: Stephen Bryson, Jon Jenkins, and Nikunj Oza; SETI: Douglas Caldwell, Jeffrey Smith, and Joseph Twicken; USRA: Miguel Martinho and Hamed Valizadegan; INTERNS: Pedro Gerum (I^2 Intern from Brazil, now Assistant Professor at Cleveland State University), Kaylie Hausknecht (Intern from Harvard University), Noa Lubin (I^2 Intern from Israel), Nikash Walia (Intern from University of Illinois at Urbana-Champaign), and Laurent Wilkens (Intern from Delft University of Technology, the Netherlands)

POINT OF CONTACT: Hamed Valizadegan, hamed.valizadegan@nasa.gov

First Gov logo
NASA Logo - nasa.gov