Machine Learning Discovers Over 10,000 Candidate Exoplanets in NASA Survey Data
Summary: Researchers using machine learning algorithms to analyze the first year of observations from NASA's Transiting Exoplanet Survey Satellite (TESS) have identified 10,091 new candidate exoplanets from a scan of 83 million faint stars—a discovery that could more than double humanity's confirmed exoplanet catalog of over 6,200 worlds.
As of today, humanity has confirmed the existence of more than 6,200 exoplanets—worlds orbiting stars beyond our solar system. But a new study published in Nature may dramatically revise that number upward. Researchers applied machine learning to TESS's first year of observational data, systematically scanning 83 million stars and discovering 10,091 previously uncatalogued candidate bodies in a single survey.
TESS detects exoplanets by identifying periodic "transits"—the subtle dimming of a star's light when a planet passes in front of it from the satellite's perspective. However, previous analysis methods focused primarily on brighter stars. The new study extended the search to stars 16 times fainter than typical TESS targets, revealing a vast population of candidates that had previously been overlooked.
All newly discovered candidates currently remain in "candidate" status pending confirmation. Some may ultimately prove to be non-planetary objects or merely noise in the data, while others could become prime targets in the search for extraterrestrial life. The research team plans to analyze additional TESS observation cycles to further expand the exoplanet database.