Normal algorithms found nothing, but using machine learning, a research team uncovered eight previously overlooked signals from space.
Los Angeles/Berkeley - "Are we alone in the universe?" is probably one of the oldest questions of mankind and at the same time a question to which there is no satisfactory answer to this day.
There is a whole branch of research for the search for intelligent life beyond Earth, called SETI (Search for Extraterrestrial Intelligence).
Researchers in this field are primarily looking for so-called "technosignatures" in space - meaning signals of technological developments in foreign civilizations that penetrate into space, as
fr.de
reports.
So far, none have been found - the famous "Wow!" signal, which was picked up once in 1977, has never been received again and has not been resolved to this day.
And other efforts have so far been unsuccessful.
For a long time, research had the problem that too little data was available. In recent years, it has been struggling with a completely new problem: huge amounts of data are now available, the analysis of which takes a lot of time.
In addition, terrestrial signals - smartphones, satellite navigation and more - repeatedly cause false positive results.
Researchers want to answer the oldest of all questions: Are we alone in the universe?
There is a reason for the flood of data: In 2015, the billionaire Yuri Milner and the astrophysicist Steven Hawking founded the “Breakthrough Lists” project – one million stars are to be searched for times when intelligent life existed in this context.
Several radio telescopes are used for this.
These look for radio signals coming from the direction of a star and constantly changing frequency.
This is how a signal should behave if its transmitter is on a planet that moves through space just like Earth.
Look into the depths of the universe - This is how "Hubble" sees the universe
Look into the depths of the universe - This is how "Hubble" sees the universe
A new study now shows that machine learning can be a useful addition in the SETI area.
"It's a new era for SETI research that's dawning thanks to machine learning technology," Franck Marchis, planetary astronomer at the SETI Institute in Mountain View, California, told
Nature
magazine .
Machine learning finds eight unknown signals from the universe
For a new study, published in the journal
Nature Astronomy
, a research team led by University of Toronto's Peter Ma applied deep learning technology to a dataset that had previously been examined.
"In total, we searched 150 terabytes of data from 820 nearby stars that were already probed using classical techniques in 2017 and in which no signals of interest were discovered at the time," explains study lead author Ma.
This time, however, the research group found something new: eight previously unidentified signals that are of interest for research.
The eight signals have not been received since then, but for the research team, the result shows one thing above all: "We believe that work like this will help speed up the detection rate," Ma said in a statement, adding, "It is about an answer to the big question 'Are we alone in the universe?'.”
Machine learning helps to search through large amounts of data
Cherry Ng, an astronomer at the SETI Institute in California and collaborator on the study, said: "These results dramatically illustrate the possibilities of applying advanced machine learning and machine vision methods to the data challenges in astronomy .” This will lead to new discoveries, Ng believes, emphasizing: “The application of these techniques on a large scale will be of great importance for the science of radio-technosignatures.”
So it may be an artificial intelligence that one day discovers the first signals of an extraterrestrial intelligence.
But maybe not: The astronomer Jean-Luc Margot from the University of California at Los Angeles (UCLA) suspects that SETI will use a mixture of classical approaches and machine learning in the future to sift through the vast amount of data.
Machine learning is “not a panacea,”
Nature
quotes the astronomer as saying.
Dan Werthimer, SETI researcher at the University of California at Berkeley, agrees: "The machines can't do everything yet." (tab)