AI and the World's Hard Problems · Astronomy & space · arXiv · 2026-10-23
Decoupling candidate dual AGN from chance superpositions in the GOTHIC survey via a deep-learning framework
Dual active galactic nuclei are hard to find in large imaging surveys, where foreground stars and chance alignments can mimic them. The authors re-examined 46,061 galaxies that an earlier pipeline had rejected, using a YOLOv11 deep-learning detector trained on annotated SDSS images. The model reached precision 0.919 and recall 0.905, yielding 29,605 candidates, of which 54.5–62% looked like genuine dual nuclei on visual inspection, though higher-resolution follow-up is needed for confirmation.