By Ilana MacDonald, Dunlap Institute for Astronomy and Astrophysics, University of Toronto

A new citizen science project has launched thanks to a team that includes researchers from the Dunlap Institute. “Galaxy Zoo: Euclid” will allow members of the public to help classify galaxies in images from the European Space Agency’s (ESA) Euclid Space Telescope. This will be the first time that the public has the chance to see the data from Euclid’s initial data set of several hundreds of thousands of galaxies.

The original version of the Galaxy Zoo Project began in 2007 using images from the Sloan Digital Sky Survey (SDSS). For the last 17 years, hundreds of thousands of citizen scientists have helped identify millions of galaxies from other telescopes, such as the Hubble Space Telescope and the James Webb Space Telescope (JWST). Classifying galaxies in this way has allowed scientists to understand their evolution, and how galaxies interact with each other and their surroundings. Much of this can be deduced from a galaxy’s shape!

The Euclid Space Telescope was launched in July 2023 and joined fellow space telescopes JWST and Gaia at Lagrange Point L2 about 1.5 million kilometers from the Earth. Its mission is to map out billions of galaxies to eventually help scientists learn about how dark matter influences the structure of the visible universe. The images returning from Euclid are of a much higher resolution than can be attained on the ground and cover a much larger area of the sky than Hubble or JWST.

Because Euclid will be sending 100GB of data back to the Earth every day for six years, astronomers must enlist the help of Artificial Intelligence (AI) to help with the huge job of labelling all that data. The team has developed an AI called Zoobot to help with the task. Dunlap Fellow Dr. Mike Walmsley is the technical lead for this project at the University of Toronto (UofT).

Using the human brain’s exceptional ability to recognize patterns, citizen scientists labelling galaxies by their shape will provide Zoobot with an excellent set of examples to teach it to accurately classify galaxies on its own. Since Euclid will be producing tens of millions of images of galaxies, it would be impossible for humans alone to take on the task of classifying them.  And so, the Zoobot AI will learn from the volunteers’ decisions and complete this colossal endeavour.

“AI models are only as good as the data they are trained on,” says Walmsley, “By teaching the Zoobot AI to classify galaxies in Euclid images, Galaxy Zoo volunteers are powering the creation of the largest and most detailed galaxy catalogue ever made.”

UofT undergraduate students Khalid Edris and Junbo Li have been spending their summer helping Zoobot learn from the classification done by citizen scientists using Galaxy Zoo: Euclid. They are working under the supervision of Dr. Walmsley and UofT Prof. Josh Speagle to select the best data for training Zoobot’s galaxy models.

“It’s amazing to see how the public’s classifications directly improve our AI model, Zoobot, and contribute to our understanding of the universe,” says Li. “Together, we’re building the most detailed galaxy catalogue ever, and that’s something truly extraordinary.”

“It’s actually quite hard to comprehend the scale of this project,” Edris adds. “We just hope that our algorithms can be the starting domino in a larger-scale project that Galaxy Zoo: Euclid is hoping to achieve by optimizing the labelling process of the millions of galaxies Euclid has captured.”

If you would like to get a sneak peek at Euclid’s newest data and help Zoobot learn how to recognize different galaxy features, visit Galaxy Zoo to get started!