(ORDO NEWS) — Ohio State University astronomers have identified about 116,000 new variable stars. These celestial bodies were discovered by the Planetary Automated Survey of Supernovae (ASAS-SN), a network of 20 telescopes around the world that can observe the entire sky about 50,000 times deeper than the human eye. Ohio State researchers have been running the project for nearly a decade.
The scientists explain how they used machine learning techniques to identify and classify variable stars – celestial objects that fade in and out of brightness over time, especially when viewed from a chosen vantage point on Earth.
In fact, even our Sun is considered a variable star.
Scientists explain that variable stars are a kind of stellar laboratory in the universe where we can study and learn more about how stars function and the little subtleties they have.
In 2018, the project moved to g-band filters, lenses that can detect more varieties of blue light, and the network went from seeing about 60 million stars at once to more than 100 million.
But unlike ASAS-SN’s civilian science campaign, which relies on volunteers to select and classify astronomical data, the new study required the help of artificial intelligence.
“If you want to look at millions of stars, then you can’t do it on your own. The process will take forever,” says Tarindu Jayasinghe, co-author of the experiment, Ohio State Astronomy Doctor. “So we had to be creative, like using machine learning techniques.”
The new study used data from Gaia, the 3D Mapping Mission of our galaxy, as well as data from 2MASS and AllWISE. The team used a machine learning algorithm to compile a list of 1.5 million variable star candidates from a catalog of about 55 million isolated stars.
The researchers then further reduced the number of candidates. Of the 1.5 million stars they studied, almost 400,000 turned out to be true variable stars. More than half of them were already known to the astronomical community, but 116,027 were new discoveries.
Although machine learning was required to complete the study, the team says there is still a role for citizen scientists to play. Volunteers under the Citizen Science Campaign have already begun to identify low-quality data, they said.
But using a training set of all this bad data allows the team to modify and improve the overall performance of the algorithm.
—
Online:
Contact us: [email protected]
Our Standards, Terms of Use: Standard Terms And Conditions.