Scientists Teach AI To Spot Far-Away Galaxy Clusters

Despite their mind-blowingly massive size (millions of lightyears across), galaxy clusters can be pretty hard to spot.

So, researchers at Lancaster University have turned to a trusted friend—artificial intelligence—for assistance.

The new “Deep-CEE” (Deep Learning for Galaxy Cluster Extraction and Evaluation) technique was developed by PhD student Matthew Chan to speed up the process.

Astronomer George Abell pioneered the sport of galaxy cluster finding in the 1950s, using a magnifying lens, photographic plates, and his own two eyes to locate the cosmic collections.

He manually analyzed some 2,000 photographic plates, detailing the astronomical coordinates of his data. That work ultimately resulted in the so-called “Abell catalogue” of rich galaxy clusters.

Now, of course, we have technology that can do in a fraction of the time what Abell took years to accomplish.

Deep-CEE builds on the researcher’s approach to identifying galaxy clusters. But instead of a human astronomer, it uses an AI model trained to “look” at color images and identify galaxy clusters.

Repeatedly shown examples of known, labelled objects in images, the algorithm learned to identify and classify galaxy clusters in images that contain many other astronomical objects. (Which is more than I can say for myself.)

Chan has already applied Deep-CEE to the multi-spectral imaging and spectroscopic Sloan Digital Sky Survey at Apache Point Observatory in New Mexico.

Moving forward, he hopes to run the model on additional equipment, including the Large Synoptic Survey Telescope—a wide-field survey reflecting telescope, currently under construction in Chile and due to come online in 2021.

By automating the discovery process, according to Lancaster University, scientists can quickly scan sets of images and return precise predictions with “minimal” human interaction.

“Data mining techniques such as deep learning will help us to analyze the enormous outputs of modern telescopes,” John Stott, Chan’s PhD supervisor, said in a statement. “We expect our method to find thousands of clusters never seen before by science.”

A paper detailing Deep-CEE has been submitted to the journal Monthly Notices of the Royal Astronomical Society, and can be found online.

More on Geek.com:

• High-Def Images Help Uncover Galaxy’s Oldest Star Cluster
• Universe’s Densest Galaxy Cluster Discovered
• Google Algorithm Detects Lung Cancer Better Than Human Doctors

The post Scientists Teach AI To Spot Far-Away Galaxy Clusters appeared first on Geek.com.

from Geek.com https://ift.tt/32fW4QY
via IFTTT