Astronomers use images of the distant universe captured by the Subaru Telescope and use artificial intelligence (AI) to locate and classify spiral galaxies.
In these ultra wide field of view images, AI has achieved very high accuracy in locating spiral galaxies. This technology, combined with citizen science, is expected to produce more discoveries in the future.
A team of astronomers, mainly from Japan’s National Astronomical Observatory (NAOJ), applied a deep learning technique (AI) to classify galaxies in large image datasets acquired by the Subaru Telescope. Due to its high sensitivity, as many as 560000 galaxies are detected in the image. It will be extremely difficult to visually process so many galaxies with the naked eye for morphological classification. AI enables the team to perform processing without manual intervention.
Since 2012, the automatic processing technology of automatic feature extraction and judgment based on deep learning algorithm has developed rapidly. Now, in terms of accuracy, they usually surpass humans and are used in self driving cars, security cameras and many other applications. Dr. Ken Ichi Tadaki, a project assistant professor at NAOJ, has come up with the idea that if AI can classify images of cats and dogs, it should be able to distinguish between “galaxies with spiral patterns” and “galaxies without spiral patterns.”. Indeed, artificial intelligence has successfully classified galactic morphology using human prepared training data with an accuracy rate of 97.5%, and then applied the trained artificial intelligence to the entire data set, which identified about 80000 spiral galaxies.
The technology has now proved to be effective and can be extended to classify galaxies into more detailed categories by training AI based on a large number of human classified galaxies.
NAOJ is now running a citizen science project, galactic cruise, in which citizens examine images of galaxies taken with the Subaru Telescope for features that suggest that the galaxy is colliding or merging with another galaxy.
Associate professor masuki Tanaka, a consultant to Galaxy cruise, has high hopes for using artificial intelligence to study galaxies, and said: “the Subaru strategic plan is serious big data, which contains almost countless galaxies. Scientifically, working with citizen astronomers and machines to process such large data. By using deep learning based on the classifications made by citizen scientists in Galaxy cruise, it is possible to find a large number of colliding and merging galaxies
Editor in charge: Tzh