NASA has discovered 301 new exoplanets using the ExoMiner deep neural network. The Kepler spacecraft found new celestial bodies, whereas the neural network helped scientist confirm their status, the space agency's blog post reads.
The transmit method is usually used to detect exoplanets. When the planet moves around its star, it happens to be in front of it at some point, changing the star's brightness, which helps scientists understand it is an exoplanet. The neural network is trained on past confirmed cases and false positives, which allows it to distinguish planets from other objects while leveraging NASA's Pleiades supercomputer.
According to preliminary data, the newly discovered exoplanets are located outside the habitable zone of their stars and are not similar to Earth.
ExoMiner did not replace the professionals. Instead, it helped those who analyze data received from the telescope. Without artificial intelligence, the task of determining which signals are coming from a planet and which not would take a long time since Kepler monitors thousands of stars with even more planets revolving around them.
Since the mid-1990s, scientists have managed to find 4,569 planets outside the solar system. Thanks to ExoMiner, this number has increased to 4,870.
Scientists now plan to modify the neural network and use it for other future missions.