Artificial Intelligence is definitely everywhere, including in space. Indeed, it is by using this one that astronomers from the University of Texas at Austin, in partnership with Google, succeeded in discovering two new planets hidden in the archives of the K2 mission of NASA's Kepler Space Telescope.
Once again, the effectiveness of AI has been demonstrated. The Universe being immense, it is not necessarily easy to find any trace of planet in our discovery of space.In an article to be published soon in The Astronomical Journal, astronomers have succeeded in detecting two new planets through the use of an algorithm using deep learning.
The latter, designed by Anne Dattilo, has thus made it possible to detect signals omitted or missed by traditional methods to detect planets. The data collected during Kepler's extended K2 mission differed considerably from those collected during the initial spacecraft mission, and Kepler constantly oscillated, and it was through the AI that the data could be used.
Other team members include NASA Sagan member at Austin University, Andrew Vanderburg, and Google engineer Christopher Shallue. Vanderburg explained how this algorithm and other AIs would play an important role in the search for Earth-like planets in the near future. "If we want to know the total number of extrasolar planets, we need to know the number of planets we have discovered, but also the number of planets that have escaped our detection. That's where AI comes in," he explains.
"Artificial intelligence will help us to analyze the data in a uniform way, even if each star was surrounded by a planet the size of the Earth, we will not find them all when we look with Kepler. We know there are many planets that we don't see for these reasons."
Very distant planets
The two new planets discovered are located in the constellation of Aquarius. Baptized K2-293b and K2-294b, they gravitate around a star located respectively 1300 and 1230 light-years from our solar system. These two planets "are typical of the planets found at K2" according to Anne Dattilo. "They are located very close to their stars, have short orbital periods, and are hot. They are also a little larger than the Earth. "
In the future, the algorithm could help to find other planets again by searching all the data of K2, or about 300,000 stars. It could also be used by Kepler's successor, the TESS, launched less than a year ago.