Humans can find exoplanets in the data of the space telescope. The image is from NASA.
Over 300 previously unknown exoplanets have been discovered by a new artificial intelligence algorithm.
NASA's first dedicated exoplanet hunter, the Kepler Space Telescope, has observed hundreds of thousands of stars in the search for potentially habitable worlds outside our solar system. Even after the telescope's demise, the calatog of potential planets continues to produce new discoveries. Human experts look at the data for signs of exoplanets. A new method called ExoMiner can now mimic that procedure and speed up the catalog search.
The telescope stopped working in November of last year, but it was still able to look for signs of a planet crossing in front of a star's disk. According to a NASA statement, not all dimmings are caused by exoplanets, and scientists had to follow elaborate procedures to distinguish false positives from real stuff.
These 10 super extreme exoplanets are out of this world.
A neural network is a type of artificial intelligence that learns and improves its abilities when fed enough data. In less than 10 years, the telescope discovered thousands of planet candidates, nearly 3000 of which have since been confirmed. That is a majority of the 4,569 exoplanets that have been discovered.
The light curve of a star is used to calculate how large a portion of the star the planet seems to be covering. They would look at how long it takes the planet to cross the star's disk. The observed brightness changes are not likely to be explained by an exoplanet. The researchers were able to add 301 previously unknown exoplanets to the catalog at once with the help of the ExoMiner algorithm.
"If the machine learning project says something is a planet, you can be sure it's a planet," said Hamed Valizadegan, the project lead and machine learning manager with the Universities Space Research Association at the NASA Ames Research Center. "ExoMiner is more reliable and accurate than both existing machine and human experts because of the biases that come with human labeling."
Scientists are looking to use ExoMiner to sift through the data from other exoplanet-searching missions, such as NASA's current TESS or the European Space Agency's Planetary Transits and Oscillations.
None of the newly confirmed exoplanets are likely candidates to host life, as they are outside of the habitable zones of their parent stars.
The paper was accepted for publication in the Astrophysical Journal, NASA said in the statement, and a draft of the paper is available to read on the preprint site arXiv.org.
Follow Tereza Pultarova on social media. Follow us on social media.
