Scientists have added a whopping 301 newly confirmed exoplanets to the total number of exoplanets.
Scientists recently added a whopping 301 recently validated exoplanets to the total number of exoplanets. The multitude of planets is the newest of 4,569 already validated planets orbiting a multitude of distant stars. All at the same time? The answer lies in a new deep neural network called ExoMiner.
Deep neural networks are machine learning methods that learn a task automatically when they are given enough information. ExoMiner is a new deep neural network that uses NASA’s supercomputer, Pleiades, to distinguish real exoplanets from various types of fraudsters or “false positives”. Its design is inspired by various tests and properties that human experts use to confirm new exoplanets, and it learns from the use of previously confirmed exoplanets and false positives.
ExoMiner complements professionals who analyze data and decipher what is a planet and what is not. In particular, data collected from NASA’s Kepler spacecraft and K2, its next mission; For missions like Kepler with thousands of stars in view, each with the potential to host multiple potential exoplanets, studying huge data sets is a time-consuming task. ExoMiner solves this dilemma.
“In contrast to other machine learning programs for exoplanet detection, ExoMiner is not a black box; there is no mystery why it decides something is a planet or not, ”said Jon Jenkins, exoplanet scientist at the Center for NASA Ames Research in California’s Silicon Valley. “We can easily explain which properties of the data cause ExoMiner to reject or approve a planet.”
What is the difference between a confirmed and a validated exoplanet? A planet is “confirmed” when different observation techniques reveal properties that can only be explained by one planet. A planet is “validated” based on statistics, that is, how likely or unlikely it is to be a planet based on the data.
In an article published in the Astrophysical Journal, the Ames team shows how ExoMiner discovered the 301 planets using data from the remaining set of possible planets or candidates in the Kepler Archives. The 301 machine-validated planets were originally discovered by Kepler Central Channeling Science Operations and promoted to planetary candidate status by Kepler’s Office of Science, but no one was able to validate them as planets until ExoMiner.
The paper also shows how ExoMiner is more accurate and consistent in eliminating false positives, and better able to reveal the real signatures of planets orbiting their parent stars, while also giving scientists a chance to see in detail what ExoMiner is about led to his conclusion.
“When ExoMiner says something is a planet, you can be sure it is a planet,” added Hamed Valizadegan, ExoMiner project leader and machine learning manager for the Space Research Association of Universities in Ames. “ExoMiner is very accurate and, in some ways, more reliable than existing machine classifiers and human experts, which it has to mimic because of the distortions that come with human labeling.”
None of the recently confirmed planets are believed to be similar to Earth or to be in the habitable zone of their parent stars, but they have similar properties to the general population of confirmed exoplanets in our galactic neighborhood.
“These 301 discoveries help us better understand the planets and solar systems beyond our own and what makes ours so unique,” said Jenkins.
While the search for more exoplanets with missions that use transit photometry, such as NASA’s Transiting Exoplanet Reconnaissance Satellite (TESS), and the European Space Agency’s upcoming PLAnetary Transits and Oscillations of Stars (PLATO) mission, ExoMiner will have more opportunities to show that you are up to the task.
“Now that we’ve trained ExoMiner on Kepler data, with some fine tuning we can transfer that knowledge to other missions, including TESS that we’re currently working on,” said Valizadegan. “There is room to grow.”
NASA Ames directed the Kepler and K2 missions for NASA’s Science Mission Directorate. JPL led the development of the Kepler Mission. Ball Aerospace and Technologies Corporation operates the flight system with support from the Laboratory for Atmospheric and Space Physics at the University of Colorado at Boulder.
