An image recognition method based on artificial intelligence can help scientists discover exoplanets. Scientists at the University of Geneva and the University of Bern described a process developed jointly with Diasitek in Astronomy and Astrophysics.
Most exoplanets have been discovered by the transit method so far. This is based on tiny fadings of starlight caused by the passage of planets orbiting their sun. But in many planetary systems, the interaction between worlds changes this repetition and makes it impossible to detect planets. This is why Swiss scientists used artificial intelligence for image recognition. The computer was taught to calculate the effect of interplanetary interactions, which made it possible to discover exoplanets – planets outside the solar system – that had hitherto been incapable of being detected by experts.
The method they have developed can also be used to detect illegal dumping on Earth, for example.
“Using a large number of examples, it is possible to teach a machine to consider all parameters and calculate the effect of interaction between planets,” reads a statement from the University of Geneva.
When the method was first applied, the researchers discovered two exoplanets, Kepler-1705b and Kepler-1705c, which could not be detected with the prior art. The planetary system thus revealed is a gold mine to gain knowledge about exoplanets. The method makes it possible to estimate the radii of planets and provides information about their mass, density, and composition.
The use of artificial intelligence, especially deep-learning AI, is becoming more widespread in astrophysics, whether in processing observational data or analyzing the results of massive computational simulations, noted Yann Alibert, a professor at the University of Bern.