Artificial intelligence found dozens of hidden planets in NASA data
Kyiv • UNN
Astronomers from the University of Warwick, with the help of AI, confirmed 118 exoplanets, including 31 new ones. The RAVEN algorithm analyzed data from 2.2 million stars.
Astronomers from the University of Warwick, using a new artificial intelligence tool, have confirmed the existence of over 100 exoplanets, including 31 previously unknown ones. To do this, scientists analyzed data from NASA's TESS satellite, which searches for planets by the faint dimming of starlight as planets pass in front of them. This is reported by Phys.org, writes UNN.
Details
The study focused on over 2.2 million stars collected during the first four years of the mission.
A new algorithm called RAVEN played a key role in the discovery. It helped filter out true planetary signals from false positives, which are often created by binary stars or other cosmic phenomena.
Using our newly developed RAVEN pipeline, we were able to verify 118 new planets and over 2000 high-quality planet candidates, almost 1000 of which are completely new.
Among the findings are very rare planets
Among the new discoveries are several particularly interesting categories: ultra-short-period planets that orbit their stars in less than a day, rare objects from the so-called "Neptunian desert", and multi-planetary systems that were previously unknown.
The challenge is to determine whether the dimming is truly caused by a planet orbiting a star, or by something else, and that's the question RAVEN tries to answer.
This could change the pace of new world discovery
Scientists believe that the new approach could seriously change the rules of the game in astronomy. If previously verifying exoplanet candidates required a lot of time and resources, now AI allows for faster and more accurate identification of real planets among vast arrays of cosmic data.
In fact, this means that dozens or even hundreds of undiscovered worlds may still be hidden in NASA's already collected archives.