CNET News Video
AI helps NASA discover two new planetsA neural network built by an astronomer and a Google software engineer was trained to spot patterns caused by planets, and it found a pair of surprise worlds in two separate star systems.
There's officially another solar system with eight planets in it. Now know that almost every star- NASA held the press conference to discuss two planets discovered using a neural network and machine learning. The event focused specifically on Kepler-90i, a planet orbitting the star Kepler-90. The discovery brought the total number of known planets in the system to eight, which ties with our very own solar system. It's the first star known to have at least as many planets orbiting it as our own sun. And because of the eight planets orbiting Kepler-90 are located in the equivalent of our inner solar system, Scientists said it could mean systems with eight or more planets are much more common than previously known. So how did a machine discover a planet? Christopher Shallue, Senior Software Engineer at Google AI, and NASA Sagan post-doctoral Fellow Andrew Vanderberg built a neural network and taught it to scan signals from existing data sets collected by the Kepler telescope. Machine learning allowed the network to flag potential candidates for planet designation, spotting them even if the signals were originally too weak to catch. The algorithm scan dips in brightness as possible planets travelled in front of their respective stars. NASA hopes the development means discovering additional information and data about existing systems by giving astronomers even more time to investigate patterns flagged as possible planets. What do you think machine learning could help astronomers accomplish? Leave your ideas down in the comments. For more, check out cnet.com. I'm Ashley Skepta, Be good humans.