Conceptual illustration of how artificial intelligence classifies different types of galaxies according to their morphologies. Credit: NAOJ / HSC-SSP
Astronomers have applied artificial intelligence (AI) to ultra-wide field-of-view images of the distant universe captured by the Subaru telescope, and have reached a very high point accuracy for finding and classifying spiral stars in those images. This technique, combined with civil science, is expected to provide further discoveries in the future.
A research group, consisting of astronomers primarily from the National Astronomical Observatory of Japan (NAOJ), applied a deep-learning technique, a kind of AI, to classify galaxies into a large dataset of images obtained with the Subaru Telescope. Due to its high sensitivity, at least 560,000 galaxies are detected in the images. It would be very difficult to visually process this large number of galaxies with human eyes for morphological classification. The AI enabled the team to perform the processing without human intervention.
Automated processing techniques for extraction and judgment of functions with deep-learning algorithms have been rapidly developed since 2012. Now they normally surpass humans in terms of accuracy and are used for autonomous cars, security cameras, and many other applications. Dr. Ken-ichi Tadaki, a project assistant professor at the NAOJ, came up with the idea that if AI could classify images of cats and dogs, they should be able to distinguish “spiral galaxies” from spiral galaxies. Indeed, with training data prepared by humans, the AI classified the galaxy morphologies with an accuracy of 97.5%. When the trained AI applied to the entire dataset, it identified spirals in about 80,000 galaxies.
Now that this technique has been proven effective, it can be expanded to classify galaxies into more detailed classes, by training the AI based on a substantial number of galaxies classified by humans. NAOJ is now running a civilian science project “GALAXY CRUISE”, where citizens are examining galaxy images taken with the Subaru Telescope to look for features that suggest the galaxy collides or fuses with another galaxy. The adviser of “GALAXY CRUISE,” Associate Professor Masayuki Tanaka has high hopes for the study of galaxies with artificial intelligence and says, “The Subaru Strategy Program is serious Big Data with an almost innumerable number of galaxies. Scientifically, it is very interesting to see such large numbers. “Taking data with a collaboration of civic astronomers and machines. By using deep learning above the classifications made by civic scientists in GALAXY CRUISE, chances are, we can find a large number of colliding and fusing galaxies.”
Reference: “Spin Parity of Spiral Galaxies II: A catalog of 80k spiral galaxies using big data from the Subaru Hyper Suprime-Cam Survey and deep learning” by Ken-ichi Tadaki, Masanori Iye, Hideya Fukumoto, Masao Hayashi, Cristian E Rusu, Rhythm Shimakawa and Tomoka Tosaki, July 2, 202, Monthly announcements from the Royal Astronomical Society.
DOI: 10.1093 / mnras / staa1880
