Detecting polar ring galaxies via deep learning

ABSTRACT Polar ring galaxies (PRGs) are peculiar galaxies that show a ring of stars, gas, and dust oriented roughly over the poles of the central ‘host’ galaxy (i.e. roughly orthogonal to the disc of the host galaxy). The formation models for these rings involve mergers or tidal interactions of the host galaxy with another galaxy. Although the identified PRGs look different from each other, they all have a ring that is not in the same plane as the disc of the host galaxy. Unlike in galaxies such as our Milky Way, where stars form in spiral arms, the rings exemplify an unusual star formation environment in PRGs. Known PRGs are a relatively small population, so expanding their sample is essential to understand their nature and evolution. Unfortunately, finding more PRGs in large galaxy surveys is a challenge. To help improve this situation, we have built and validated a deep learning model pipeline to automatically distinguish PRGs from other galaxies. Our model achieved a classification ROC-AUC of almost $99{{\ \rm per\ cent}}$. We also achieved an average recall score of $96\, {{\ \rm per\ cent}}$ for the PRG class in the test data set. From these successful results, this work shows that our deep learning pipeline can help to discover new PRGs in future large surveys.