Student Harnesses AI to Create Algorithm for Analyzing 1.5 Million Celestial Objects

This is the first paid job of Mattéo Paz, who finished his secondary school studies in parallel.

“” “The model that I have implemented can be used for other studies in the time domain in astronomy, and potentially for everything that is in temporal form. I see an interest in the analysis of graphics (scholarship holders), where information is also presented in the form of chronological series and where the periodic components can be crucial. We could also study atmospheric effects such as pollution, where day-free seasons and cycles play a crucial role ', explained the student to the media Phys.org.

A mentor to support the student

Mattéo Paz took six weeks to develop his algorithm. To do this, he is helped by his mentor, the astronomer and the main scientist of the IPAC, Davy Kirkpatrick. “” “It allowed me to live an experience of limitless learning. I think it is thanks to him that I have progressed as much as a scientist ”, explained Mattéo.

“” “If I see their potential, I want to make sure they exploit it. I will do my best to help them ”, commented Davy Kirkpatrick. The two men turned to Neowise (Near-Earth Object Wide-Field Infrared Survey Explorer), a former NASA infrared telescope. His role was to seek asteroids for 10 years, but he was also able to identify the heat variations in celestial objects.

Variable objects analyzed by algorithm

For astronomers, these variable objects are difficult to detect. In addition, the data on variable objects were not used by the Neowise team. “” “My idea for summer was therefore to take a small piece of sky and see if we could find variable stars. We could then present them to the astronomical community saying: “Here are new manual discoveries; imagine the potential of this data set.” “, explained Mattéo Paz's mentor.

But Mattéo Paz did not intend to exploit all the data manually … hence the creation of his algorithm. He relied on his mentor in parallel with his work on the algorithm in order to have additional information on the most relevant astronomical and astrophysical aspects.

This AI model developed and perfected by Mattéo Paz is now able to exploit “Raw data from Neowise's observations ” But also to analyze the results, according to Phys.org. Thus, 1.5 million new potential objects have been identified and classified and a complete catalog of objects whose brightness varied considerably will be potentially published by the two scientists in 2025.