Portuguese 
English 
Spanish 
4 min of reading 
0 comments 
Astronomy Lives on Surprises, Many of Which Come from Where You Least Expect It. Among Giant Telescopes and Billion-Dollar Missions, a High School Student Managed to Capture the Attention of the Scientific Community. Using Artificial Intelligence, He Transformed Forgotten Data into Discoveries That Reveal How Much More There Is to Explore in the Universe.
A high school student from California surprised the scientific community by using artificial intelligence to detect over 1.5 million previously unidentified space objects.
The peer-reviewed work was published in the The Astronomical Journal, instantly lending credibility to the discovery.
An AI Pipeline Created by a Teenager
Matteo Paz, a resident of Pasadena, joined Caltech’s Planet Finder Academy in 2022, a program that offers young students real experiences with astronomical challenges. There, he received guidance from Davy Kirkpatrick, a scientist at the Infrared Processing and Analysis Center (IPAC).
-
Butterfly robots, giant bees, and even a mechanical serpent are attracting attention in Germany and show how industry is copying nature to create faster and smarter machines.
-
Goodbye to critical metals? Scientists create a new revolutionary aluminum that could slash industrial costs, accelerate the energy transition, and completely transform global technology in the coming years.
-
Giant superatoms promise to eliminate the biggest problem in quantum computing and could surprisingly accelerate the arrival of powerful universal machines.
-
While drought threatens millions, China is advancing with a technology capable of transforming seawater into drinking water for less than US$1, generating green hydrogen, and converting waste into energy and valuable products, in a solution that can redefine the future of water scarcity in the world.
He gained access to data from NASA’s NEOWISE telescope, launched in 2009 to monitor near-Earth asteroids. With over a decade of infrared observations of the entire sky, the data set contained information about billions of light sources, many still unexamined.
The problem was the scale. The dataset exceeded 200 billion lines of observations. The team initially considered studying only a fraction manually. Paz, however, decided to seek another solution.
Technology Applied to Astronomy
With experience in mathematics, programming, and time series analysis, the student developed a machine learning pipeline to identify subtle signals hidden in the data. In just six weeks, he managed to build a system capable of recognizing weak and variable light sources.
According to Kirkpatrick, the results emerged almost immediately. As the model was refined, the patterns became clearer.
Identification of Variable Phenomena
The great achievement was detecting objects that flicker, pulse, or decrease in intensity. This type of behavior can indicate the presence of quasars, eclipsing binary stars, or even supernovae.
Furthermore, the algorithm revealed signals that had gone unnoticed in previous analyses. Some variables changed so slowly or so quickly that they did not fit into conventional studies.
The Meeting Between Big Data and Big Sky
To achieve these results, Paz applied techniques such as Fourier transforms and wavelet analysis. These tools allow the observation of variations in signals over time and were essential for dealing with NEOWISE’s sampling limitations.
For months, the student worked alongside scientists like Shoubaneh Hemmati, Daniel Masters, Ashish Mahabal, and Matthew Graham. Together, they applied the model across the entire database.
The effort resulted in a catalog with over 1.5 million variable sources, now documented in a scientific paper. The public release of the catalog is scheduled for 2025.
This material is expected to support future observations with advanced telescopes, such as the Vera Rubin Observatory and the James Webb Space Telescope, enabling new studies on stellar cycles and distant cosmic phenomena.
A Path Under Construction
The discovery changed Paz’s routine. Even while still in high school, he became a paid research assistant at IPAC, continuing to develop his pipeline and training new colleagues at the academy.
His performance stands out because he utilized skills typically seen at the graduate level. Among them are time series modeling and the use of algorithms for interpreting astronomical data.
All of this was made possible thanks to the Mathematics Academy of the Pasadena Unified School District, a rigorous public program that offers training beyond the traditional curriculum.
Potentials Beyond Astronomy
Although it was created to analyze spatial data, the pipeline has applications in other areas. Any dataset involving temporal variations can be studied with the same logic.
Therefore, fields like finance, environmental monitoring, and even neuroscience can benefit from the approach. Small fluctuations in data series often contain critical information that eludes traditional analysis.
This versatility demonstrates how methods developed for astronomy can be adapted to various sectors. It is a clear example of how interdisciplinary science grows supported by machine learning.
Recognition and Support
According to Kirkpatrick, supporting talented young people is essential. He emphasized that he makes it a priority to ensure that promising students have the necessary conditions to achieve their goals.
Paz stated that he sees the work as a starting point. He believes that the pipeline can be expanded, leading to discoveries both in space and in areas on Earth.
A Student Who Challenged the Limits
The journey showcases how curiosity and access to opportunities can lead to unexpected advancements. What started as a summer challenge transformed into a massive catalog of information about the universe.
And the phrase he chose to sum up the achievement explains the impact of the accomplishment: “I mapped the invisible.”
Be the first to react!