New planet on distant star discovered via artificial intelligence

The planet is 30 percent larger than Earth, and significantly hotter.

ByABC News
December 14, 2017, 5:55 PM

— -- The discovery of a new planet orbiting a distant star means there is another solar system a lot like ours.

That’s according to NASA researchers who announced the discovery of Kepler-90i, the eighth planet circling the Kepler-90 star. The planet is 30% larger than Earth, and significantly hotter with temperatures believed to exceed 800 degrees Fahrenheit.

“The Kepler-90 star system is like a mini version of our solar system,” says Andrew Vanderburg, a NASA Sagan Postdoctoral Fellow and astronomer at the University of Texas at Austin. “You have small planets inside and big planets outside, but everything is scrunched in much closer.”

The Kepler-90 star is 2,545 light years from Earth. The furthest of the eight planets is similar to the distance between Earth and our Sun.

The planet was discovered using data from NASA’s Kepler Space Telescope. Vanderburg and his colleague Christopher Shallue trained a computer to look for changes in brightness of the Kepler-90 star. Brightness dips as planets in front of their star. The computer was able to sift through past data to discover the previously unknown eighth planet.

Machine learning developed by Google was crucial to the discovery. In an approach to artificial intelligence, computers ‘learn’ the inconsistencies to look for that could signal the existence of a new planet.

“In my spare time, I started Googling for ‘finding exoplanets with large data sets’ and found out about the Kepler mission and the huge data set available,” said Shallue, a senior software engineer with Google’s research team Google AI. “Machine learning really shines in situations where there is so much data that humans can’t search it for themselves.”

NASA said that in four years they have seen 35,000 possible planetary signals from the Kepler system. Only the most promising signals can be analyzed by humans, but a machine-learning computer can look at every signal.

“We got lots of false positives of planets, but also potentially more real planets,” said Vanderburg. “It’s like sifting through rocks to find jewels. If you have a finer sieve then you will catch more rocks but you might catch more jewels, as well.”