Dec. 14 (UPI) -- NASA scientists have found a planetary system with as many planets as our own.
"Scientists have found for the first time eight planets in a distant planetary system," Paul Hertz, astrophysics division director at NASA Headquarters, said during a teleconference on Thursday that was live-streamed on NASA TV.
Astronomers were aware of seven of the eight planets orbiting the Kepler 90 star. The discovery of the new planet, Kepler-90i, was made possible by machine learning.
"Machine learning is a way to teach computers to recognize patterns," said Christopher Shallue, senior software engineer at Google AI. "The key idea is to let the computer learn by example."
Shallue and his colleagues used a specific type of machine learning called neural networking to process the Kepler-90 data. Neural networks mimic the way the brain processes signals and learns to recognize patterns.
The network is organized by layers. Signals on an initial layer trigger patterns on another. The many layers form a network capable of recognizing a diversity of patterns.
Using already verified Kepler data, scientists trained the network to identify the patterns created by the transits of exoplanets across the face of a distant star.
Astronomers were already using machine learning to identify promising signals. But their previous methods identified only potential exoplanet candidates among strong signals, which were then analyzed by scientists.
The new neural network is able to do the much more difficult and tedious work of finding weak signals among much larger datasets.
"The key contribution of the machine learning is that it was able to search much larger amount of signals than humans would have been able to search in a reasonable amount of time," Shallue said.
Kepler-90 is a sun-like star located in the constellation Draco, some 2,545 light-years from Earth. Kepler-90i, the newly discovered planet, orbits the star once every 14.4 days.
"The new planet is small enough that it is probably rocky and doesn't have a thick atmosphere," said Andrew Vanderburg, astronomer and a NASA Sagan postdoctoral fellow at the University of Texas. "But Kepler 90i is probably not a place I'd like to go visit. The surface is likely scorching hot."
Scientists estimate the planet's surface temps exceed 800 degrees Fahrenheit, similar to the extreme temperatures measured on Mercury.
The new neural network also found an extra planet in the Kepler-80 system, a sixth planet. Kepler-80g is an Earth-sized world whose presence accounts for the orbital resonance that keeps it and its tightly-packed neighbors from colliding and collapsing.
Compared to our own solar system, both the Kepler-80 and Kepler-90 planets are highly concentrated around their host stars.
"The Kepler-90 star system is like a mini version of our solar system. You have small planets inside and big planets outside, but everything is scrunched in much closer," said Vanderburg.
Searching for planets with more distant orbits is difficult because potential transit signals happen less frequently.
Whether via machine learning or the eyes of an astronomer, finding exoplanets requires a pattern, and a pattern requires repetition. If a planet only crosses the face of its host star once every two years, it can take a long time for a recognizable pattern to form in the Kepler data.
"It's very possible that Kepler 90 has even more planets that we don't know about yet," Vanderburg said. "It would almost be surprising to me if there weren't any more planets around this star."
But as Kepler and its successors continue to field more and more data over longer time spans, it's likely new patterns -- and new exoplanets -- will be revealed.
"We've only begun to scratch the surface," Vanderburg said.
Until then, astronomers will use the new neural network to parse datasets already collected by Kepler and do the kind of analysis that identified the presence of Kepler-90i and Kepler-80g.
The details of the duo's discovery will soon be published in the Astronomical Journal.
"These results demonstrate the enduring value of Kepler's mission," said Jessie Dotson, Kepler's project scientist at NASA's Ames Research Center in California. "New ways of looking at the data -- such as this early-stage research to apply machine learning algorithms -- promises to continue to yield significant advances in our understanding of planetary systems around other stars."