Japanese astronomers have developed a new artificial intelligence technique!

Artificial Intelligence

Due to random variations in galaxy shapes to remove noise in astronomical data, Japanese astronomers have developed a new artificial intelligence (AI) technique. Extensive testing and training were done on large mock data created by supercomputer simulations; the new tool was applied to actual data from Japan’s Subaru Telescope. They found out that from using this method, the mass distribution derived is consistent with the currently accepted models of the Universe. For analyzing big bad, this a powerful tool from planned and current astronomy surveys.

To study the large-scale structure of the Universe, wide-area survey data can be used through measurements of gravitational lensing patterns. The gravity of a foreground object in gravitational lensing can distort the image of a background object, such as a more distant galaxy. The large-scale structure, which mainly consists of mysterious dark matter, can distort the shapes of distant galaxies. In an area averaging over many galaxies is required to create a map of foreground dark matter distributions.

Advertisement

But this method of looking at many galaxies images runs into a problem. It is challenging to identify whether a galaxy image is distorted by gravitational lensing or the galaxy is distorted. This called shape noise, and this one of the limitations the research has, which is studying the large-scale structure of the Universe.

To compensate for shape noise, a team of Japanese astronomers, to compensate for shape noise 1st used ATERUI II to generate 25,000 mock galaxy catalogues based on accurate data from the Subaru Telescope. ATERUI II is the world’s most powerful supercomputer dedicated to astronomy. Then realist noise was added by them to these perfectly known artificial data sets, and from the mock data to recover the lensing dark matter trained an AI. The AI was able to recover after training previously unobservable fine details, helping to improve the understanding of cosmic dark matter. On accurate data covering 21 square degrees of the sky, using this AI, the team found out that the distribution of foreground mass is consistent with the standard cosmological model.