High school students use AI to reveal 1.5 million unknown objects in the universe

Through his research at Caltech, local high school students revealed 1.5 million previously unknown objects in space, expanding the possibilities of the NASA mission and publishing a single authored paper.

Matteo (Matthew) Paz’s article published in The Astronomical Journal describes the new AI algorithms that led to these discoveries, allowing other astronomers and astrophysicists to adapt to their own research.

Pass wanted to learn more about astronomy as he brought him to a Stargaze lecture published in California when his mother was in elementary school. In the summer of 2022, he came to campus to learn astronomy and related computer science at the Caltech Planet Finder Academy, led by Professor Andrew Howard of Astronomy.

Astronomer and senior IPAC scientist, Davy Kirkpatrick, served as pass leader.

“I’m very lucky to see Davey,” Pass says. “I remember the first day I spoke to him. I said I was considering working on paper for it to come out. This is a much bigger goal than six weeks. He didn’t disappoint me. I think he’s the reason I’ve grown as a scientist.”

Kirkpatrick grew up in the Tennessee farming community and realized his dream of becoming an astronomer with the help of 9th grade chemistry and physics teacher Marilyn Morrison. She told him and his mother that he had a chance and explained the courses to prepare for college.

“I wanted to give the same kind of mentoring to someone else. “If I see their potential, I want to make sure they’re reaching it. I’ll do whatever I can to help them.”

Kirkpatrick also wanted to gather more insights from Neowise (Infrared Survey Explorer of the Neowise), a now-retired infrared telescope that had been scanning the sky for more than a decade in search of asteroids and other objects near Earth.

The NASA telescope was busy observing asteroids, but it also detected various heat in other distant space objects that flashed violently, pulsing, or dimmed. Astronomers call these variable objects. It’s a difficult phenomenon to catch, like a quasar, an explosive star, and an attractive star to each other.

However, data on these variable objects has not yet been utilized. If the Neowise team can identify these objects and make them available to the astronomical community, the resulting catalogue can provide insight into how space entities will change over the years.

“At that point, we were creeping up towards 200 billion rows on the table of all the detections we’ve done over a decade,” says Kirkpatrick. “So my idea of ​​summer was to take a small part of the sky and see if we could find different stars. Then we could highlight them in the astronomical community, saying, “Here are some new things we discovered. Imagine what the possibilities of the dataset are.”

Path was not intended to manually sift the data. His studies prepared him to bring a new perspective to the challenge. He was interested in AI during electives with integrated coding, theoretical computer science and formal mathematics.

Path knew that AI would train most with the vast and orderly datasets that Kirkpatrick gave him. Puzz had the advanced mathematics knowledge necessary to enjoy programming. He was already studying advanced undergraduate mathematics at the Mathematics Academy in the Pasadena Unified School District.

Therefore, PAZ set out to develop machine learning techniques to analyze the entire data set and analyze the entire potential variable object. In these six weeks he began drafting the AI ​​model and began to show some promises. While he worked, he consulted with Kirkpatrick to study related astronomy and astrophysics.

“Every meeting with Davy is 10% of the job, and 90% of the US just chats,” Paz says. “It was so cool just to have someone talk about that kind of science.”

Kirkpatrick linked Paz with Caltech astronomers Shoubaneh Hemmati, Daniel Masters, Ashish Mahabal and Matthew Graham. Paz and Kirkpatrick learned that the particular rhythm of Neowise’s observations mean that many objects that flashed once immediately or that were gradually changing over time cannot be systematically detected and classified.

Just as summer was over, there was still much work to do. In 2024, Pass and Kirkpatrick joined forces again, this time Puzz coached other high school students.

Currently, PAZ is improving its AI model to process all raw data from Neowise’s observations and analyze the results. The algorithm trained to detect micro-differences in telescope infrared measurements was categorized by flagging 1.5 million potential new objects in the data. In 2025, Paz and Kirkpatrick will publish a complete catalog of objects with significantly different brightness in Neowise data.

“The models I have implemented can be used for other time domain studies in astronomy and potentially other in time format,” says Paz. “We could have relate to (stock market) chart analysis. Here, information can be chronological and regular components. We can also study atmospheric effects such as pollution, where regular seasons and day and night cycles play a major role.”

Currently, Paz is an employee of California while he is graduating from high school. He works for IPAC’s Kirkpatrick, managing, processing, archive and analyzing data for Neowise and several other NASA and NSF supported space missions. It’s Puzz’s first pay job.