Conceptual illustration to show how artificial intelligence classifies various types of galaxies in the Universe according to their morphology. The base-line techniques are applied in many fields, such as classification of animal images. Application of convolutional neural networks to images of galaxies is expected to yield morphological classification of the galaxies into not only "spiral galaxies" but also other types of galaxies, including "barred spiral galaxies" and "merging galaxies." (Credit: NAOJ/HSC-SSP)
A group of astronomers have used state of the art artificial intelligence algorithms to classify more than half a million galaxies in images obtained with the Subaru Telescope located on the slopes of Maunakea. About 100 years ago, the American astronomer Edwin Hubble discovered that various types of galaxies exist in our Universe, from beautiful spiral galaxies to smooth elliptical-shaped galaxies. A research group from the National Astronomical Observatory of Japan have now succeeded in building a completely automated way of identifying different types of galaxies according to their morphologies, without the need for any human inspection of the images. In particular, they applied a “deep-learning” technique, which is one of the fundamental techniques of artificial intelligence, to a large set of galaxy images obtained over 300 nights of observing with the Subaru Telescope. This work will help us understand the different populations of galaxies and how they evolve over time, which is still one of the biggest enigmas in galactic astronomy today.
Find out more in the Subaru Press release, complete with video tour!