Galaxy Morphology in the deep learning era

The galaxy population today is known to present a bimodal distribution of shapes, going from disky, rotationally supported star-forming galaxies to spheroidal, pressure supported quiescent galaxies. The origins of this galaxy diversity are still debated and understanding the physical process that build the galaxy population is the main purpose of the field of galaxy formation and evolution. In order to progress in this front, it is desirable to have large samples of galaxies with labeled morphologies at different cosmic epochs. This is a field where deep learning has had a major impact by enabling the automated classification of galaxies with unprecedented accuracy. We review some of the key results obtained by our group in this paper.