British researchers have identified 50 new planets using AI technology, using an algorithm, studying old NASA files.
His astronomers and computer scientists University of Warwick, created one algorithm machine learning with AI technology, to research old data containing NASA containing thousands of potential candidate planets.
It is not always clear which of the candidate planets are real. When scientists look for exoplanets, that is, planets outside our solar system systemic, looking for transitions of light that indicate the passage of a planet. Of course, this could also be caused by other factors, such as background interference or even errors in camera.
So here comes the new AI technology, which can tell the difference and find the planets.
The research team trained the algorithm with the data collected by the NASA Space Telescope Kepler, which spent nine years in deep space in a "global hunting mission ", and has now retired. Once the algorithm learned to accurately separate real planets, it was used to analyze old data sets that had not yet been confirmed. It was there that he found them 50 exoplanets.
The 50 exoplanets, which revolve around other stars, may be as large as Poseidon or even smaller than Earth, the university said in a press release. The trajectory of some of them lasts up to 200 days, and while others can be as short as one day. And now that, thanks to AI technology, astronomers know that planets are real, they can prioritize them for further observation.
The findings were published last week in Monthly Notices of the Royal Astronomical Society.
«In terms of planetary verification, no one has used a machine learning technique in the past,Said in the press release o David Armstrong of the University of Warwick, one of the leading authors of the study.
"Machine learning has been used to rank candidate planets, but never in a probabilistic context, which is what it takes to validate a planet."
Now that researchers know it works, they hope to use AI technology for current and future missions telescopes, or even discover and study other planets.
Η research It also states that the algorithm will be able to "identify thousands of planets" that scientists have never seen.
«We still have time to devote to algorithm training, but once that is done, it will be much easier to apply it to future candidates., »Said Armstrong.
Armstrong added that the algorithm could be used to analyze data from Transiting Exoplanet Satellite Survey (TESS) of NASA, a sky exploration program, which completed its main mission on July 4.
Mapping about the 75% of the sky, TESS he found 66 new confirmed exoplanets and almost 2.100 potential candidates. Among the confirmed exoplanets, one found the size of Earth and possibly habitable, orbiting a 100-year-old light star. He also discovered a planet orbiting two suns, like Tatooine in the Star Wars movies.
TESS is now on an extended mission until 2022, as scientists work on AI technology to validate and confirm which of the other potential candidate planets are real.