The aim of this project is to use recent advances in active learning to efficiently construct a dataset to understand galaxy and black hole co-evolution. Since expert astronomer annotation is the major bottleneck to discovering enough examples, this project builds upon the successes of citizen science projects to construct a catalogue of morphologically classified active galactic nuclei. By managing the exploration-exploitation trade-off, active learning is a machine learning approach that uses annotator effort more efficiently.
In addition to a significant improvement in understanding of the physics of co-evolution, we expect that the same software suite can be re-used for other scientific questions. This means that the developed software will be a key part of future astronomical surveys.