“I Mapped the Invisible”: American High School Student Dazzles Astronomers by Finding 1.5 Million Hidden Cosmic Objects
Indian Defence Review • • science
Key Points:
- A 17-year-old high school student, Matteo Paz, developed an AI model named VARnet to analyze NASA’s NEOWISE infrared telescope data, which had been unexamined for over a decade due to its massive size of nearly 200 billion brightness readings.
- Paz's model efficiently processes light curves by combining wavelet decomposition, a customized Fourier transform, and a convolutional neural network to classify celestial objects into four categories, achieving high accuracy with a processing time of under 53 microseconds per source.
- VARnet identified 1.5 million candidate variable objects in the NEOWISE data, which require further astronomical verification; the full catalog is expected to be published in 2025, offering a comprehensive dataset for studying infrared variability across the sky.
- Paz’s work earned him the $250,000 first prize in the 2025 Regeneron Science Talent Search, and he continues to contribute to astronomy research and mentor other students while finishing high school at Caltech.
- The AI model has potential applications beyond astronomy, including other fields involving time series data, such as chart analysis, demonstrating the broader impact of Paz’s research.