A pair of astronomers from the European Space Agency (ESA) discovered more than 800 previously undocumented “astrophysical anomalies” hidden in Hubble’s archives. To do this, researchers David O’Ryan and Pablo Gomez trained an AI model to scour Hubble’s 35-year-old dataset, look for strange objects, and flag them for manual review. “This is a wealth of data in which astrophysical anomalies can be found,” O’Ryan said. statement.
Studying space is difficult. There’s a lot of it, it’s noisy, and the flood of data generated by instruments like the Hubble Space Telescope can overwhelm even large research teams. And sometimes space is weird. Very strange. Enter AI, which is very good at sorting through large amounts of information and detecting patterns – flagging quirks that astronomers might otherwise miss.
The model used by astronomers, called AnomalyMatch, scanned approximately 100 million image cutouts from the Hubble Legacy Archive, marking the first time the dataset has been systematically searched for anomalies. Think oddly shaped galaxies, light distorted by the gravity of massive objects, or planet-forming disks visible edge-on. AnomalyMatch took only two and a half days to sift through the dataset, which is much faster than a human research team would attempt to do the task.
conclusion, published in the journal astronomy and astrophysicsAbout 1,400 “anomalous objects” were detected, most of which were merging or interacting galaxies. Other anomalies include gravitational lenses (light distorted into circles or arcs by massive objects in front of them), jellyfish galaxies (which have hanging “tentacles” of gas), and galaxies with large clusters of stars. “Perhaps most interestingly, there were several dozen objects that completely defied classification,” ESA said in a report. blog post.
“This is a brilliant use of AI to maximize the scientific output of the Hubble collection,” Gomez said. “Finding so many anomalous objects in the Hubble data, where you might expect many had already been found, is a fantastic result. It also shows how useful this tool will be for other large datasets.”
