For over ten years, NASA’s NEOWISE telescope surveyed the sky in infrared, primarily searching for near-Earth asteroids. This mission gathered an astonishing archive of nearly 200 billion data points, capturing not just asteroids, but also distant stars, quasars, and galaxies. However, deciphering this treasure trove wasn’t easy. Astronomers knew it held gems like flickering quasars and dimming stars, but processing that massive amount of data was a daunting task.
### The Young Innovator
In 2023, a 17-year-old named Matteo Paz entered the picture. A Pasadena High School student passionate about astronomy from an early age, he joined Caltech’s research program. Matteo was no ordinary teenager; he had breezed through advanced math classes and had a background in machine learning. During his first day at Caltech, he proposed an ambitious idea: not just to analyze a small section of the data, but to develop a model that could scrutinize the entire NEOWISE dataset.
Dr. Davy Kirkpatrick, his mentor, saw potential in Matteo. Growing up in a Tennessee farming community, he understood the impact of guidance and mentorship. He was determined to support Matteo in reaching his goals.
### The Power of Machine Learning
Matteo’s model, named VARnet, is a machine learning algorithm that processes data through three main phases. First, it minimizes background noise caused by cosmic rays and instrument errors. Next, it identifies periodic features from the data. Finally, a convolutional neural network classifies each observation into categories such as variable stars or transient events.
Recent studies show that machine learning is revolutionizing fields like astronomy. For example, a 2023 survey by the American Astronomical Society found that 80% of astronomers believe machine learning will significantly impact future discoveries.
### Uncovering Hidden Treasures
When Matteo applied VARnet to the NEOWISE dataset, it flagged 1.5 million potential variable objects. This impressive number doesn’t mean that all these candidates are new discoveries. Instead, each one represents a lead that astronomers will need to investigate further.
Matteo collaborated with Caltech experts to refine the machine-learning techniques. The insights gained showed that NEOWISE’s observational patterns limited its ability to detect certain fleeting phenomena. This collaboration not only improved the accuracy of the model but also demonstrated the importance of teamwork in scientific research.
### Future Possibilities
The complete catalog of variable objects is set to be published in 2025. This dataset will open new avenues for statistical studies on infrared variability across the entire sky. Matteo believes VARnet has the potential for broader applications beyond astronomy, suggesting it might benefit fields such as finance, where similar time series analysis is crucial.
Matteo’s achievements stand out, especially since he won a $250,000 prize in the prestigious Regeneron Science Talent Search. For Dr. Kirkpatrick, this wasn’t just a win for Matteo; it represented a triumph in mentoring and unlocking young talent.
In a world increasingly driven by data and technology, Matteo’s story is a powerful reminder of the potential within our youth and the exciting horizons that await in scientific research.
Source link
