Research results that AI is useful for detecting ``signals emitted by aliens with intelligent civilization''
Curtin University , Australia, has announced research results on `` Attempts to detect signals emitted by aliens using AI ''. .
Earthlings are not the only ones building highly intelligent civilizations in this universe. , researchers are analyzing observational data to look for signs of intelligent life. A research team led by Mr. Danny Price, who is a senior researcher at the International Radio Astronomy Research Center at
A deep-learning search for technosignatures from 820 nearby stars | Nature Astronomy
AI is helping us search for intelligent alien life – and we've found 8 strange new signals
Price has made tremendous progress in the AI field over the past few years, with AIs of incredible ability emerging, such as image-generating AIs Midjourney and Stable Diffusion, and text-generating AIs ChatGPT. pointed out. “AI is now being used in almost every scientific field to help researchers with their routine classification tasks. It has also been helpful in clinical trials, with promising results so far.”
Radio astronomers looking for signs of intelligent extraterrestrial life use data from radio telescopes that observe the faint radio waves emitted by celestial bodies to identify signals (technosignatures) that cannot be produced by natural astrophysical processes. is detected. However, the vast amount of data generated by radio telescopes contains a large amount of radio interference emitted from telephones, Wi-Fi, artificial satellites, etc., and the detection of technosignatures is like searching for a needle in a cosmic haystack. It is said that it is a thing.
To solve this problem, Peter Ma, a student at the University of Toronto , Canada and the first author of the paper, devised a method to quickly detect technosignatures using AI. In the first step, Ma injects simulated technosignatures into real data and uses this dataset to train an AI to identify technosignature features.
In the second step, a random forest classifier , a type of machine learning algorithm, was used to determine whether a particular signal was noteworthy or due to radio interference.
Green Bank Telescope in the United States were input to AI, AI detected 20,515 notable signals. This result seems to reduce the number of candidate signals by two orders of magnitude compared to previous analyzes using the same dataset.
After training the AI algorithm, when more than 150 terabytes (equivalent to 480 observation times) of data observed by the
Further manual classification revealed that eight of the signals had techno-signature features rather than radio interference. However, although they re-observed with a radio telescope to verify these signals, they said that they could not re-observe any of the signals in tracking observations. 'The most likely explanation is that these signals were anomalous signs of radio interference, not aliens,' Price said.
Unfortunately, the problem of radio interference persists, but astronomers are working to mitigate it. The radio telescope ' MeerKAT ' in the Republic of South Africa has 64 parabolic antennas with a diameter of 13.5m and functions as one radio telescope. False positives can be greatly reduced. The research team led by Price recently deployed a powerful signal processor on MeerKAT.
'If astronomers succeed in detecting a technosignature that cannot be explained by radio interference, it would strongly suggest that humans are not the only creators of technology in the galaxy,' Price said. If nothing is detected, that doesn't mean we're the only 'intelligent' species with technology, it just means we're not looking for the right kind of signal. Or it could be that our telescopes aren't sensitive enough to detect faint communications from distant exoplanets.'
Regarding AI, Mr. Price said, ``AI algorithms do not 'understand' or 'think'.AI is excellent at pattern recognition and has been proven to be very useful in tasks such as classification, but the problem It has no ability to solve. AI only does the specific tasks it is trained to perform.' 'Our research uses AI to classify signals as either radio interference or genuine technosignature candidates.' We created an algorithm that does, and our algorithm performs better than we expected.'