Project Title: Machine-Learning Segmentation of Galaxy Mergers Using Volunteer Classifications
"Galaxy mergers help us understand how galaxies evolve over time, which in turn helps us understand how the universe itself changes throughout its history."
— Vivasvaan Aditya Raj
Published July 28, 2026
What if thousands of volunteers and artificial intelligence could work together to uncover some of the universe’s most hidden cosmic collisions?
Galaxy mergers occur when galaxies interact and eventually combine, shaping how they grow and evolve. But not every interaction leads to a merger. Some galaxies simply pass by one another in what's known as a "flyby," continuing on separate paths after a minor interaction. Whether galaxies merge or just interact briefly, the faint traces they leave behind can be difficult to detect, and even experts don’t always agree on what they’re seeing.
Vivasvaan Aditya Raj, runner-up in the Physical Sciences and Engineering category at this year’s Research Computing Exhibition, is developing a machine-learning system to improve how these mergers are identified. His work combines citizen science with machine learning to detect the subtle signatures of galaxy interactions more consistently.
The project uses data from Galaxy Zoo, a citizen science platform where volunteers classify images of galaxies. Their contributions provide the training data that allows machine learning models to learn what mergers look like in real astronomical observations.
A key challenge is scale. Building reliable models requires thousands of carefully reviewed volunteer classifications before training can begin, making data collection time-intensive even though the computing itself is fast.The project relies heavily on the Minnesota Supercomputing Institute, whose high-performance computing systems make it possible to train models on tens of thousands of galaxy images in a fraction of the time it would take on a personal computer.
Looking ahead, Vivasvaan hopes this work will help build large catalogs of galaxy mergers from upcoming surveys like Euclid, enabling new studies of how galaxies interact and evolve over billions of years.
By combining citizen science, machine learning, and advanced computing, this research shows how collaboration between people and technology is helping reveal how the universe changes at its largest scales.