An Artificial Intelligence program that was initially used to identify faces on Facebook has been developed using deep learning techniques to locate and identify galaxies in deep space. This has given rise to the creation of an AI bot named ClaRAN that can scan images taken by radio telescopes to spot radio galaxies that emit powerful radio jets from massive black holes located within their centre.
Big data specialist, Dr. Chen Wu, and astronomer Dr. Ivy Wong, from the University of Western Australia node of the International Centre for Radio Astronomy Research (ICRAR) developed ClaRAN.
Black holes are found at the center of most of the galaxies and these supermassive black holes occasionally burp out jets that can be seen with a radio telescope according to Dr. Wong. These jets can stretch a long way from their host galaxies over time making it difficult for traditional computer programs to figure out where these galaxies are. This is the learning curve ClaRAN has to achieve.
The idea was mooted and developed out of an open source version of Microsoft and Facebook’s object detection and identification software. Dr. Wu says that the program was completely overhauled, trained and fed with data to recognize galaxies instead of people. ClaRAN is also an open source. She also added that ClaRAN was an example of a new paradigm called ‘programming 2.0’.
The software is based on an internal architecture and a neural network trained and fed by big data obtained from the data sourced through a highly accurate catalog from Radio Galaxy Zoo volunteers which uses citizen scientists to spot galaxies. Quite early on Wong had capitalized on the Radio Galaxy Zoo project to get data on galaxies.
With the right classification of the pattern and recognition methods, the machine can be trained to understand the behavior of each galaxy. This data optimizes the parameters inside the network to extract the desired results.
The upcoming EMU survey using the Australian Square Kilometer Array Pathfinder (ASKAP) telescope is expected to observe up to 70 million galaxies across the history of the Universe.
According to Dr. Wong, traditional computer algorithms were able to correctly identify 90 percent of the sources, which leaves 10 percent, or seven million ‘difficult’ galaxies that have to be eyeballed by a human due to the complexity of their extended structures.
Dr. Wong said ClaRAN had huge implications for how telescope observations were processed. The benefits of science can be maximized only if advanced methods can be implemented for the next generation surveys. She believes that there is no point using 40-year-old methods on brand new data, as science is trying to probe further into the Universe than ever before.
Deep learning which is the most popular form of machine learning which ClaRAN makes good use of, offers models that promise to open new windows into our universe. Researchers and dedicated volunteers can help realize the full potential of science to see some of the greatest accomplishments of our times.