A team of researchers used machine learning tools to confirm evidence of a previously unknown planet outside our solar system. With this, they demonstrated that machine learning could correctly predict whether an exoplanet is present by looking within the gas around newly formed stars.
According to the researchers, the study published in The Astrophysical Journal represents a first step towards using machine learning to identify exoplanets that may have been overlooked previously.
As data sets grow larger and larger, they become more difficult for small teams of researchers to analyze. Scientists often turn to complex machine-learning algorithms, but these can’t yet replace human intuition and our brains’ superb pattern-recognition skills. However, a combination of the two could be a perfect team. Astronomers recently tested a machine-learning algorithm that used information from citizen-scientist volunteers to identify exoplanets in data from NASA’s Transiting Exoplanet Survey Satellite .
Discovering alien signals with AI
As scientists searching for evidence of intelligent life beyond Earth, we have built an AI system that beats classical algorithms in signal detection tasks. Our AI was trained to search through data from radio telescopes for signals that couldn’t be generated by natural astrophysical processes.
Not so intelligent
AI algorithms do not “understand” or “think”. They do excel at pattern recognition, and have proven exceedingly useful for tasks such as classification—but they don’t have the ability to problem solve. They only do the specific tasks they were trained to do.
According to Cassandra Hall, until now, machine learning was only used to find forming exoplanets that were already known. Now, the researchers have proved that models can be used to make brand-new discoveries.
Some planets are harder to find than others, too. Long-period planets orbit their star less frequently, meaning a longer period of time between dips in the light. TESS only studies each patch of sky for a month at a time, so for these planets may only capture one transit instead of several periodic changes.
For the algorithm to perform accurately, though, it needs a lot of this labeled training data. “It’s difficult to get labels on this scale without the help of citizen scientists,” Nora Eisner, an astronomer at the Flatiron Institute in New York City and co-author on the study, told Space.com.
To try and verify these signals, we went back to the telescope to re-observe all eight signals of interest. Unfortunately, we were not able to re-detect any of them in our follow-up observations.
We’ve been in similar situations before. In 2020 we detected a signal that turned out to be pernicious radio interference. While we will monitor these eight new candidates, the most likely explanation is they were unusual manifestations of radio interference: not aliens
“These technologies are very important, especially for big data sets and especially in the exoplanet field,” Giada Arney, an astrobiologist at NASA’s Goddard Space Flight Center in Greenbelt, Maryland, told .