Artificial intelligence helps classify galaxies

(ORDO NEWS) — Astronomers have applied artificial intelligence to the processing of images of the sky taken with the Subaru Telescope’s ultra-wide field of view camera, and have been able to achieve very high accuracy in detecting and classifying spiral galaxies.

This method, used in conjunction with projects involving the involvement of amateur astronomers in image processing, has great prospects for the future, scientists say.

The research team, led by Ken-ichi Tadaki of the National Astronomical Observatory of Japan, used deep machine learning, a form of artificial intelligence, to classify galaxies that are part of a large Subaru telescope. a collection of great scientific importance.

Due to the high sensitivity of the new method, about 560,000 galaxies were identified in the images. “Manual” classification of such a large number of galaxies by morphological features would be a very difficult task. Artificial intelligence allowed processing without human intervention.

Previously, artificial intelligence has already successfully proven itself in the classification of galaxies into “spiral” and “non-spiral”. In early research, Dr. Tadaki “trained” artificial intelligence on a set of galaxies correctly identified by amateur astronomers and applied it to another set of 80,000 galaxies that was successfully computerized with a 97.5 percent accuracy for spiral galaxies. …

Now, after receiving confirmation of the effectiveness of his method, Tadaki used it to classify galaxies into subclasses in more detail, while training artificial intelligence on a much larger set of galaxies classified by the forces of amateur astronomers as “colliding” and “not experiencing” in within the Galaxy Cruise project, which is led by the National Astronomical Observatory of Japan. Tadaki believes that this “union” of machine and man promises to bring good results in the future.

Online:

Contact us: [email protected]

Our Standards, Terms of Use: Standard Terms And Conditions.

Adblock Detected

Please consider supporting us by disabling your ad blocker