An international team of astronomers validated more than 100 hidden planets outside the solar system using an artificial intelligence tool called RAVEN. It was developed at the University of Warwick (United Kingdom).
The system analyzed data from NASA’s Transiting Exoplanet Survey Satellite (TESS), reviewing observations of 2.2 million stars collected over four years.
Among the findings are:
- 31 previously unidentified planets.
- More than 2,000 additional high-quality candidates.
- Ultra-short period exoplanets, with orbits less than 24 hours.
- Planets in the “Neptunian desert”, a region where their presence is uncommon according to theoretical models.
Validation and precise characterization
Dr. Marina Lafarga Magro, lead researcher, explained that they managed to validate 118 new planets through “one of the best-characterized samples of planets near Earth.” This will allow for selecting systems for more detailed analysis.
The study not only confirmed the existence of the planets but also characterized them, identifying multiplanetary systems with bodies not previously detected.
How RAVEN works
Dr. Andreas Hadjigeorghiou, lead developer, noted that the challenge lies in distinguishing whether the dimming of starlight is due to a planet or other phenomena, such as eclipsing binary stars. RAVEN was trained with a catalog of simulated planets and astrophysical phenomena. This allowed for identifying complex patterns and differentiating true exoplanets from false positives.
Professor David Armstrong, co-author of the study, highlighted that the technology enables a systematic and objective analysis of large volumes of astronomical data. This reduces uncertainty in calculations up to ten times compared to previous studies.
Statistical results
A complementary study revealed that between 9 and 10% of solar-type stars host short-orbit exoplanets, a value consistent with observations by the Kepler mission. Additionally, the first direct measurement of the presence of planets in the “Neptunian desert” was achieved, concluding that only 0.08% of solar stars analyzed host this type of body.
Tools open to the community
The researchers made public catalogs and interactive tools available for other specialists to select targets of interest for future observations, including ground-based telescopes and upcoming space missions like the European Space Agency’s (ESA) PLATO.
The massive and precise validation of exoplanets through artificial intelligence opens new possibilities for astronomical exploration. This integrative approach not only allows for the discovery of hidden worlds but also refines cataloging methodologies and statistical analysis, accelerating knowledge about the distribution and characteristics of planetary systems in the galaxy.
