By Stef Verhagen
A team of researchers from the SETI Institute, Breakthrough Listen, and global institutions has used deep learning on a dataset of 820 nearby stars and found eight new signals of interest.
With the help of improved artificial intelligence, the search for technosignatures may speed up significantly. - Image Credit: Kevin Key via Shutterstock / HDR tune by Universal-Sci
Despite many years of research, definitive proof of extraterrestrial life has yet to be found. Nevertheless, many scientists believe that extraterrestrial life exists, and some have even suggested that intelligent life might exist in our own galaxy.
However, the lack of evidence for extraterrestrial life has left many wondering why we haven't found it yet. One of the reasons may be that, as of yet, we have only searched a microscopic part of the milky way due to, among other reasons, the limitations of outdated algorithms and a limited search area in the galaxy.
In order to speed things up, University of Toronto undergrad Peter Ma came up with a project that uses the help of artificial intelligence in the form of machine learning to search available data for signals of interest.
How is machine learning used in the search for alien life?
Machine learning is a branch in the field of research regarding artificial intelligence. It focuses on developing algorithms and statistical models that enable computers to perform tasks that typically require human intelligence, such as improving from experience, recognizing patterns, and making predictions.
In the case of the search for extraterrestrial life, machine learning can be used in a number of ways. For example, using it to analyze enormous sets of data collected from astronomical observations. Researchers can use algorithms and models that are trained on previous observations and simulations, allowing for more accurate and efficient analysis. That way, it can pick up on anomalies and identify patterns that could indicate the presence of extraterrestrial life.
Another way machine learning can help is by identifying and classifying exoplanets in data from telescopes, such as the famous James Webb Space Telescope. Algorithms can be used to identify the characteristics of exoplanets, such as their size, composition and atmosphere, which can be used to determine the probability of extraterrestrial life existing on those exoplanets.
Both of these methods aim at different types of extraterrestrial life. Searching for favorable conditions for life on exoplanets and analyzing their atmosphere for signs will more likely lead to the discovery of microbial life or animal/plant life than intelligent life. On the other hand, searching for evidence of life by detecting technosignatures among the stars means looking for intelligent life. Unsurprisingly given the name (search for extraterrestrial intelligence), SETI is focused on the latter search method.
Interesting article: Why extraterrestrial intelligence is more likely to be artificial than biological (Universal-Sci)
Searching nearby stars
Peter Ma and his team searched through an enormous dataset of 150 terabytes taken with the Green Bank Telescope, which contained data from 820 nearby stars. Scientists had already looked at the set six years ago (using conventional techniques), and it was labeled as 'devoid of interesting signals.'
The Green Bank Telescope, (the world's largest fully steerable radio telescope), located in West Virginia - Image Credit: John M. Chase via Shutterstock / HDR tune by Universal-Sci
The objective was to improve the classical search algorithm by using advanced deep learning techniques for quicker and more precise results. After executing the new algorithm and conducting a manual review of the data to verify the findings, the newly detected signals displayed several distinct features.
Deviating signals
In a press release, SETI shared what made the newly discovered deviating signals special. The abnormal signals had narrow spectral widths (just a few hertz), whereas signals created by natural phenomena tend to be broadband.
The signals exhibited non-zero drift rates, meaning they displayed a slope. This type of slope could suggest that the source of the signal had a relative movement in relation to the receiving equipment, making it unlikely that the source is located near the radio observatory.
The signals were present in the "ON-source" observations but not in the "OFF-source" observations. If a signal is coming from a specific celestial source, it will be visible when the telescope is aimed at the target but will disappear when the telescope is pointed elsewhere. (In contrast, human-made radio interference typically shows up in both the ON and OFF observations, indicating that the source is located nearby.)
Seen here are waterfall plots of the 8 deviating signals. Each panel has a width of 2,800 Hz and the x-axis is referenced to the center of the signal snippet where the signal is found, as indicated in column 3 of Table 1. Credit SETI Institute - (click on image to enlarge)
Scaling up
According to astronomer Dr. Cherry Ng, one of the advisors for SETI, the results show the incredible power of using modern machine-learning techniques in the field of astronomy. Scaling up the machine learning efforts may transform the field of radio technosignature science.
Currently, the team is scaling its search effort to 1 million stars and beyond, using data from the South African MeerKAT radio telescope.
Peter Ma: "We believe that work like this will help accelerate the rate we're able to make discoveries in our grand effort to answer the question 'are we alone in the universe?"
The team published their findings in a paper published in the peer-reviewed science journal nature astronomy. It is listed below this article for those interested in more details about the study.
Sources and further reading:
A deep-learning search for technosignatures from 820 nearby stars (nature astronomy)
Alien life is out there, but our theories are probably steering us away from it (Universal-Sci)
Why extraterrestrial intelligence is more likely to be artificial than biological (Universal-Sci)
James Webb Space Telescope (NASA)
Green Bank Telesecope (GBT)
SETI Institute (SETI)
Too busy to follow science news during the week? - Consider subscribing to our (free) newsletter - (Universal-Sci Weekly) - and get the 5 most interesting science articles of the week in your inbox
FEATURED ARTICLES:
The way you prepare your vegetables affects their nutritional value. What is the healthiest method to prepare them?