Abstract:
Red geysers are an important class of low star forming galaxies that show telltale signs of AGN maintenance mode driven feedback mechanism that keeps them quenched. There are many questions left to be answered about the nature of these low luminosity AGNs in maintaining quenched state - such as which stage of maintenance is dominant in the local galactic population In order to answer these questions, one requires a statistically significant sample of red geysers to conduct a study of the distribution of their properties. From a sample of ∼ 4700 in an earlier data release of the SDSS-MaNGA survey, 139 red geysers were identified by manually inspecting each example. The latest data release of ∼ 10000 galaxies has not been scoured for red geysers yet. Our goal is to build an automated machine learning model to solve this problem. We present our results with
different models and discuss a novel algorithm based on the few shot learning paradigm that can perform the task with ∼ 99% accuracy.
