The most popular and comprehensive Open Source ECM platform
Let us know more
Artificial Intelligence: AI Speeds the Identification of New Exoplanets
We know about eight or possibly nine planets in our solar system orbiting our Sun. Planets outside the solar system that orbit around other stars are called exoplanets. Astrophysicists have been on a search to identify exoplanets, looking for planets that have an atmosphere and environment that resembles the one we have on earth. More than 4500 exoplanets have been discovered.
Astronomers use telescopes to observe and detect exoplanets, and with the rapid advance of technology, the speed and our ability to identify new exoplanets has increased dramatically. NASA launched the Kepler spacecraft in 2009 to look for exoplanets. It’s goal was to help identify exoplanets closer to the size of earth. Prior to Kepler, the exoplanets discovered were typically very large planets because observing movement of the planet around a star was difficult to observe with smaller planets.
The Kepler Space Telescope stopped working in November 2018, but during it’s lifetime, it collected large amounts about hundreds of thousands of stars in attempt to identify exoplanets. The data collected by Kepler is now being processed using AI to help spot and pick out astronomic details.
NASA Ames is using an AI algorithm that they’ve named ExoMiner to help identify exoplanets using data from Kepler and other sources. Hamed Valizadegan, ExoMiner project lead, said that “when ExoMiner says something is a planet, you can be sure it’s a planet. It is highly accurate and in some ways more reliable than both existing machine classifiers and the human experts it’s meant to emulate because of the biases that come with human labeling.”
Yann Alibert, Professor at the University of Bern, said that “the use of AI, in particular of ‘deep learning’ as in this paper, is becoming increasingly widespread in astrophysics, whether to process observational data, as we did here, or to analyze the results of gigantic numerical simulations producing terabytes of data. What we have developed in this study is a new example of the fantastic contribution that these techniques can make to our field, and probably to all fields of research.”
