It’s said there are more stars in the universe than there are grains of sand in all the earth’s deserts and beaches—10 times more, by some estimates. So, imagine the plight of planet hunters, who must sort through images of countless stellar bodies to find telltale images of alien worlds. However, by using neural-network-based technology and image recognition techniques, researchers have identified new planets whose existence had previously been overlooked, a development that could alter the science of planet hunting.
The researchers have identified two new planets—technically known as exoplanets—that are orbiting distant stars called Kepler-90 and Kepler-80, each of which is located more than 1,000 light years from earth.
The discovery was made by analyzing data generated by NASA’s Kepler Space Telescope. Kepler detects planets by observing the slight dimming in the light from a star that occurs when a planet passes in front of it. The telescope monitors 145,000 stars. So far, Kepler has identified 2,341 confirmed exoplanets and 4,496 potential exoplanets, NASA said.
To find these celestial bodies, Kepler must collect astronomical quantities of data. The dataset gathered by Kepler includes a total of 35,000 possible planetary signals.
NASA uses a combination of automated and manual techniques to review this massive amount of data. However, the agency is aware that these methods sometimes fail to detect the faintest of all planetary signals.
To uncover these elusive signs hiding amidst the massive quantities of information two researchers, Christopher Shallue and Andrew Vanderburg, joined forces to apply neural-network technology developed by Google to the task of performing image recognition based on Kepler’s data.
The neural network sifted through Kepler’s data and recognized the characteristic pattern of weak transit signals from a previously-missed eighth planet orbiting Kepler-90, in the constellation Draco, according to NASA.
In the Kepler-80 system, the researchers found a sixth planet, the Earth-sized Kepler-80g.
Shallue and Vanderburg plan to expand the application of their neural network to Kepler’s full set of more than 150,000 stars.
“These results demonstrate the enduring value of Kepler’s mission,” said Jessie Dotson, Kepler’s project scientist at NASA’s Ames Research Center in Mountain View, 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. I’m sure there are more firsts in the data waiting for people to find them.”
Tyler Schulze is vice president, strategy & development at Veritone. He serves as general manager for developer partnerships, cognitive engine ecosystem, and media ingestion for the Veritone platform. Learn more about our platform and join the Veritone developer ecosystem today.