Classification of astronomical objects is not so simple. With sources at nearly unimaginable distances, it is all too common for researchers to be unable to distinguish between objects. starsOr galaxies, or quasars, or supernovae, for example. To help solve this problem, scientists have developed an artificial intelligence algorithm capable of determining the nature of cosmic sources.
Andrew Humphrey, a researcher at the Institute for Astrophysics and Space Sciences and from the University of Porto, said in Portugal.
publicity celebrity
She was a master’s degree and doctoral advisor to Pedro Cunha, a researcher at the same institutions and creator of a machine learning algorithm called SHEEP, whose study is described in an article published in the journal. Astronomy and astrophysics.
SHEEP is a supervised machine learning pipeline that estimates optical aberrations and uses this information when subsequently classifying sources as a galaxy, quasar, or star. “Optical information is the easiest to obtain, and therefore, it is very important to make a first analysis of the nature of the sources observed,” says Cunha.
In a program, a pipeline consists of a series of processing elements – processes, steps, and functions – arranged so that the output of each element is the input of the next.
“A new step in our pipeline is that before performing the classification, SHEEP first estimates optical aberrations, which are then put into the dataset as an additional resource to train the classification model,” the author explained.
Artificial intelligence trained on a 3D map of the universe
According to the team, including the redshift and coordinates of the objects allowed the AI to understand them within a 3D map of the object. Universeand used this along with color information to make better estimates of source properties.
For example, SHEEP knows there is a higher chance of finding stars closer to the plane of the Milky Way than to the galactic poles. “When we allowed AI to get a 3D view of the universe, it really improved its ability to make accurate decisions about what each celestial body was,” Humphrey said.
As highlighted by the site Physlarge-scale surveys, both terrestrial and satellite, such as Sloan Digital Sky Survey (SDSS), it produced large amounts of data, revolutionizing the field of astronomy.
Read more:
Future research, based on data from the Vera C. Rubin Observatory, the Dark Energy Spectroscopy instrument (DESI), the European Euclid space mission or James Webb Space Telescope You will continue to give us more detailed pictures.
However, analyzing all the data using traditional methods can be time consuming. Artificial intelligence will be a critical factor in analyzing and making the best scientific use of this new data.
“One of the most exciting parts is seeing how machine learning helps us better understand the universe,” Cunha said. “Our methodology shows us a possible path, while new ones are being created along the way. It is an exciting time for astronomy.”
Have you seen our new videos on Youtube? Subscribe to our channel!
