Astronomers aren't the only ones who can spot new planets. A Google-developed AI can, too.
On Thursday, Google and NASA announced the discovery of two new planets outside our solar system that were found using the company's machine learning technology.
The two planets—both over 1,000 light-years away—were glimpsed when a Google-backed computer model analyzed data taken from the Kepler Space Telescope, which was launched in 2009.
Using Kepler, astronomers have so far found over 2,300 planets via automated software and manual analysis. However, astronomers haven't spotted all the hidden gems in the telescope's data, which is actually quite overwhelming even for computers.
Kepler has been finding new worlds by measuring the brightness of a star when an orbiting planet passes across its light. For four years, the telescope has observed 200,000 stars, taking a picture every 30 minutes, to find these changes in star brightness.
Any dips in the light may indicate the presence of a circling planet. However, Kepler has recorded so much data that it translates into about 2 quadrillion (1015) possible planet orbits, Google said in a blog post.
So to streamline the planet-hunting, Google AI researcher Chris Shallue and University of Texas at Austin astrophysicist Andrew Vanderburg developed a computer model using machine learning, a subset of artificial intelligence that can involve pattern recognition. The Google-developed computing software distinguished between planets and non-planets based on the minuscule brightness changes recorded in Kepler's telescope data.
Real Life. Real News. Real Voices
Help us tell more of the stories that matterBecome a founding member
During the testing phase, the computer model proved adept at spotting the patterns; it correctly identified planets 96 percent of the time against previously vetted data. The AI-driven model was then unleashed to hunt for new planets in the data by looking at a mere 670 stars.
Lo and behold, the model found two new worlds hiding in the data. "Just as we expected, there are exciting discoveries lurking in our archived Kepler data, waiting for the right tool or technology to unearth them," Paul Hertz, a NASA director, said in a statement.
The first world, Kepler-90i, turn outs to be the eighth discovered planet in a star system over 2,500 light-years away. NASA says it the first time an 8-planet star system has been found outside our own. However, the planet itself is quite hot, and probably wouldn't be suitable for life.
The second world, called Kepler-80g, is located in a star system 1,100 light years away. Kepler-80g is actually an Earth-sized planet.
It isn't the first time machine learning has been used to look at the Kepler data, but the technology demonstrates how effective it can be at making new discoveries.
The two researchers, Shallue and Vanderburg, intend to use their computer model to go beyond 670 stars, and examine the rest of Kepler's telescope data for other planets.
Subscribe to the newsletter news
We hate SPAM and promise to keep your email address safe