[ad_1]
The researchers reused a Facebook AI that recognizes people in the photos to identify galaxies in the depths.
The new robot, named ClaRAN, analyzes the radio telescope images in the hope of locating radio galaxies that emit powerful radio jets from a supermassive black hole (SMBH).
ClaRAN is an original idea of Chen Wu, big data specialist, and astronomer Ivy Wong, both of the University of Western Australia node of the International Center for Research in Radioastronomy ( ICRAR).
Supermassive black holes exist in the center of almost all massive galaxies known to date. In the case of our Milky Way, the SMBH corresponds to the location of Sagittarius A *.
According to Wong, these black holes "burst" on occasion, jets that can be seen with a radio telescope.
"Over time, the jets can move away from their host galaxies, making it difficult for traditional computer programs to determine where the galaxy is," she said. "That's what we try to teach ClaRAN."
Based on an open-source version of Microsoft and Facebook's object detection software, the revised program is designed to recognize galaxies rather than humans.
ClaRAN is also open source and publicly available on GitHub.
By combining data from different telescopes, ClaRAN's level of "confidence" in detections and classifications is increased. Indicated by a number above the detection box, a confidence index of 1.00 indicates that ClaRAN is extremely confident that the detected source is a radio-galaxy system and that it has correctly classified (via Chen Wu and Ivy Wong / ICRAR / UWA)
We currently know about 2.5 million radio sources, but we plan to discover another 70 with the next survey on the Evolutionary Map of the Universe (UEM).
EMU is a large project that will use the new ASKAP (Square Kilometer Array Pathfinder) telescope to identify radio sources in the sky. Traditional computer algorithms must be able to correctly identify 90% of these sources.
"This leaves another 10% or 7 million" difficult "galaxies that need to be closely monitored by a human because of the complexity of their extensive structures," said Wong.
"If ClaRAN reduces the number of sources requiring a visual classification of 1%, it leaves more time for our citizen scientists to search for new types of galaxies," she added.
Wong had previously exploited the power of people to spot galaxies as part of Radio Galaxy Zoo's much-needed project.
The group volunteers helped produce the catalog used to form ClaRAN – an example of a new Wu paradigm called "2.0 programming".
"All you do is configure an extensive network, give it a ton of data and let it understand how to adjust its internal connections to generate the expected result," he said. "It's the future of programming."
A research article on ClaRAN was published today in the newspaper Monthly Notices from the Royal Astronomical Society, published by Oxford University Press.
More cosmic cover on Geek.com:
[ad_2]
Source link
