Astronomers in England say they've taught a computer to use an artificial intelligence algorithm to recognize and categorize galaxies in astronomical images.

The process, using a form of AI known as unsupervised machine learning, can allow an automatic classification of galaxies at high speed, the researchers say.

"The important thing about our algorithm is that we have not told the machine what to look for in the images, but instead taught it how to 'see,'" says Alex Hocking, a graduate student at the University of Hertfordshire.

To test their technique, the researchers applied images from the Hubble Space Telescope's "Frontier Field" project and pictures of distant clusters with several different types of galaxy, explains James Geach, Hocking's supervisor and fellow researcher.

"A human looking at these images can intuitively pick out and instinctively classify different types of object[s] without being given any additional information," he says. "We have taught a machine to do the same thing."

The algorithm was "trained" on one image, then was applied to another image where it successfully separated and classified image features a human would associate with different galaxies, such as early and late type, the researchers report in their study submitted to the Monthly Notices of the Royal Astronomical Society.

The algorithm could take on a task previously performed by thousands of volunteer citizen scientists taking part in astronomical projects such as Galaxy Zoo.

The computer algorithm could be the perfect tool to use on planned future astronomical imaging surveys where no human, or even groups of humans, would be able to adequately scan every bit of data, Geach says.

In addition to astronomy, the algorithm could have a number of other potential uses, and the team is investigating such applications, he says.

Such uses could include medicine, where the algorithm could scan thousands of images looking for tumors, or in the security field, where it might examine airport scans looking for suspicious items or people, the researchers say.

They presented the results of their work at the National Astronomy Meeting in Wales.

ⓒ 2021 TECHTIMES.com All rights reserved. Do not reproduce without permission.