
AI Uncovers Hundreds of Hidden Cosmic Anomalies in Hubble Data
A neural network named AnomalyMatch scanned nearly 100 million image cutouts from the Hubble Legacy Archive in about 2.5 days, flagging roughly 1,400 candidate anomalies. Human researchers reviewed the top results and confirmed more than 1,300 as true anomalies, including over 800 objects not previously documented, such as merging galaxies, gravitational lenses, jellyfish galaxies, and distant edge-on disks, showcasing AI's power to accelerate discovery in massive astronomical datasets.












