قالب وردپرس درنا توس
Home / Science / Artificial Intelligence Bot trained to detect galaxies

Artificial Intelligence Bot trained to detect galaxies



Fourteen radio galaxy predictions made by ClaRAN during its scan of radio and infrared data. All predictions were made with a high confidence level, displayed as a number above the detection field. A security of 1.00 indicates that ClaRAN is very confident that the detected source is a radio galaxy jet system and has classified it correctly. Credit: Dr. med. Chen Wu and Dr. Ivy Wong, ICRAR / UWA.

Researchers taught an artificial intelligence program that identifies faces on Facebook to identify galaxies in space.

The result is a KI-Bot called ClaRAN, which scans images taken by radio telescopes.

Its mission is to detect radio galaxies – galaxies emitting massive radio jets from supermassive black holes in their centers. ClaRAN is the Brainchild of big data specialist dr. Chen Wu and the astronomer Dr. Ivy Wong, both from the University of Western Australia node of the International Center for Radio Astronomy Research (ICRAR).

Dr. Wong said black holes are at the center of most, if not all galaxies.

"These supermassive black holes occasionally spit out jets that can be seen with a radio telescope," she said.

The jets can extend far from their host galaxies, making it difficult for traditional computer programs to find out where the galaxy is.

"That's what we're trying to teach ClaRAN."

Dr. Wu ClaRAN evolved from an open source version of the object recognition software from Microsoft and Facebook.

He said the program has been completely redesigned and trained to detect galaxies instead of humans.

ClaRAN itself is also open source and publicly available on GitHub

Combining data from different telescopes increases ClaRAN's confidence in its detection and classification. The number above the detection box is 1.00 dicates. ClaRAN is very confident that the detected source is a radio galaxy jet system and that it has been correctly classified. On the left is a radio galaxy jet system, which is captured by ClaRAN exclusively with data from radio telescopes. ClaRAN is not sure what to see here. There are two predictions, one for the entire system with a low reliability of 0.53 and one for the top jet only with a reliability of 0.67. On the right is the same galaxy, but overlaid with infrared telescope data. With the inclusion of data from infrared telescopes, ClaRAN has increased confidence in detection to its highest value of 1.0, and ClaRAN now includes the entire system in its only prediction. Credit: Dr. med. Chen Wu and Dr. Ivy Wong, ICRAR / UWA

Dr. Wong said the upcoming EMU survey with the WA-based ASKAP (Australian Square Kilometer Array Pathfinder) telescope is expected to see as many as 70 million galaxies in the universe's history.

She said that traditional computer algorithms 90 can correctly identify percent of sources.

"There are still 10 percent, or seven million," difficult "galaxies that must be observed by a human due to the complexity of their extended structures," Dr. Wong.

DR. Wong has previously used the power of citizen science to locate galaxies through the Radio Galaxy Zoo project.

"When ClaRAN reduces the number of sources that require visual classification to one percent, it means more time for our citizen scientists to look at new types of galaxies," she said.

A high-accuracy catalog created by volunteers from Radio Galaxy Zoo was used to train ClaRAN on how to find where the jets came from.

Dr. Wu said ClaRAN is an example of a new paradigm called & # 39; programming 2.0 & # 39 ;.

"You just have to build a huge neural network, give it a lot of data, and figure out how to shut down its internal connections to achieve the expected result," he said.

ClaRAN views more than 500 different views of radio galaxy data to make its discoveries and classifications. After searching through the different views, ClaRAN then also considers the data from infrared telescopes to refine the predictions and obtain the final recognition and classification results of a radio galaxy jet system. Credit: Dr. med. Chen Wu and Dr. Ivy Wong, ICRAR / UWA.

"The new generation of programmers spend 99 percent of their time creating records of the highest quality, and then train the AI ​​algorithms to optimize the rest.

" This is the future of programming. "

Dr. Wong said ClaRAN has great implications for the processing of telescope observations.

" If we can use these advanced methods for our next generation of surveys, we can maximize science from them, "she said. 19659005] "There are no 40-year-old methods to prove with brand-new data because we try to explore the universe further than ever before.

A research paper on ClaRAN was published today in Monthly Notices of the Royal Astronomical Society .


Explore further:
Astronomers recognize signs of supermassive black hole fusions

Magazine Reference:
Monthly announcements by the Royal Astronomical Society

Provided by:
International Center for Radio Astronomy Research


Source link