Astronomers working with the European Space Agency have used artificial intelligence to uncover a large number of unusual objects hidden deep inside decades of space data collected by the Hubble Space Telescope.
By training an AI system to scan through massive image archives, the researchers were able to find strange and rare cosmic objects that had never been properly documented before.
The research team, led by astronomers David O’Ryan and Pablo Gómez, created an AI model called AnomalyMatch to examine Hubble’s full 35 years of observations. Space research generates enormous amounts of data, and manually reviewing every image is nearly impossible, even for large teams. The AI was designed to solve this problem by quickly identifying objects that look unusual compared to normal stars and galaxies.
In just two and a half days, the AI system analyzed almost 100 million small image cutouts from the Hubble Legacy Archive. This marked the first time the entire dataset had been systematically searched for unusual or unexpected objects. Doing the same task by hand would have taken humans many years, if it could be done at all.
The results were impressive. The AI flagged nearly 1,400 anomalous objects, most of which turned out to be galaxies in the process of merging or interacting with each other. The findings also included rare gravitational lenses, where light bends into arcs or rings due to massive objects in front of it, as well as jellyfish galaxies that appear to have long streams of gas trailing behind them. Some galaxies with unusually large clusters of stars were also identified.
The discoveries were published in the scientific journal Astronomy and Astrophysics and highlight how useful artificial intelligence can be in modern astronomy. By helping scientists find strange and interesting objects faster, AI is making it easier to explore the universe and uncover hidden details that may lead to new discoveries about how space works.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
