NASA’s Kepler data added 301 planets, thanks to machine learning
Machine learning (ML) methods not only complement existing technology but also put scientific research ahead. Now, a new deep learning method has added a whopping 301 alien planets to the tally. These planets were added to the 4,569 confirmed planets orbiting several distant stars. The additions were made with the help of a deep neural network called ExoMiner, which works for NASA’s Pleiades supercomputer to detect new planets. Once provided with enough data, ExoMiner will learn the task of distinguishing between real planets and “false planets”. It is designed on the basis of various tests and characteristics that human experts use to detect alien planets. It is also provided with a database of confirmed planets and false positives.
In a paper published in the Astrophysical Journal, the team at the Ames Research Center in California’s Silicon Valley show how ExoMiner discovered 301 planets using data available in the Archives. NASA’s Kepler.
Jon Jenkins, an exoplanet scientist at the Ames Research Center, speak, “Unlike other exoplanet detection machine learning programs, ExoMiner is not a black box – it’s no mystery why it decides something is a planet or not. ” ExoMiner is transparent about data, confirming or denying a planet. A planet is confirmed using identifiable characteristics and then it is confirmed using statistics. None of the 301 newly discovered planets have Earth-like habitable conditions.
Hamed Valizadegan, ExoMiner project lead and machine learning manager, said, “ExoMiner is highly accurate and in some ways more reliable than both existing machine classifiers and the human experts it means. is for simulation because of the biases that come with human labeling.”
The researchers believe that ExoMiner has enough “room to grow.”