Artificial intelligence has invaded astronomy and the search for intelligent life in the universe, or SETI as communicated by a research at Breakthrough Listen (a science-based program to search for intelligent extraterrestrial communications in the Universe) operating from the Astronomy Department at the University of California, Berkeley. In a recent research, a SETI project led by the University of California, Berkeley, has deployed machine learning to discover 72 new fast radio bursts from a mysterious source some 3 billion light years from Earth.
So what are fast radio bursts? They are bright pulses of radio emission mere milliseconds in duration, which are assumed to originate from distant galaxies. The source of these emissions is, however, still unclear. There have been many theories ranging from highly magnetized neutron stars blasted by gas streams from a nearby supermassive black hole. Another theory suggests that the burst properties are marked with signatures of technology developed by an advanced civilization.
The director of the Berkeley SETI Research Center and principal investigator for Breakthrough Listen adds, “SETI work is exciting not just because it assists to understand in detail the dynamic behavior of fast radio bursts, but also because of the future promise holds to use machine learning for detecting signals missed by classical algorithms.”
The University of Berkeley backed Breakthrough Listen has additionally applied the successful machine learning algorithm to find new kinds of signals that may be emitting from extra-terrestrial civilizations.
Tracking Cosmic Signals
The research has pointed out that while most fast radio bursts are irregular occurrences, the source under study, FRB 121102, is unique as it is emitting repeated bursts. This has drawn the attention of many astronomers who aim to point the cause and the extreme physics involved in these regular and fast radio bursts.
The AI algorithms took the radio signals from the data that was recorded by the Green Bank Telescope in West Virginia over a five-hour period on Aug. 26, 2017. According to Berkeley SETI postdoctoral researcher Vishal Gajjar, previous analysis of 400 terabytes of data used standard computer algorithms to identify 21 bursts during the same period. The cosmic radio bursts were seen within one hour, which pointed out that the source alternated between periods of idleness and frenzied activity.
Another UC Berkeley Ph.D. student Gerry Zhang and collaborators have developed a new, powerful machine learning algorithm to reanalyze the data collected in 2017, to find an additional 72 bursts which were not detected earlier. These new findings bring the total number of detected bursts from FRB 121102 to around 300 since its discovery in 2012.
Analysing Radio Bursts
Undertaking his research Gerry Zhang said, “This work is only the beginning of using these powerful methods to find radio transients and we hope our success may inspire other serious endeavors in applying machine learning to radio astronomy.”
Zhang’s team have used some of the same techniques which are used by the internet technology companies to optimize search results and image classifications. The team has trained an algorithm referred as the convolutional neural network to point bursts found by the classical search method which was used traditionally by SETI postdoctoral researcher Vishal Gajjar and collaborators, to set it loose on the dataset with an aim to locate bursts that the classical approach might have missed.
The latest results have put new constraints on the periodicity of the pulses from FRB 121102, pointing that the pulses are not received following a regular pattern if the period of the pattern is longer than 10 milliseconds.
Whether or not FRBs themselves finally turn out to be signatures of extra-terrestrial technology, Breakthrough Listen will assist to push the frontiers to understand and unveil the secrets of the universe around the mankind.