Some galaxies blaze across telescope images, but the ones that Burçin Mutlu-Pakdil hunts are almost impossible to spot.
These dwarf galaxies are among the smallest, faintest objects in the universe, and Mutlu-Pakdil, an assistant professor of physics and astronomy, is determined to find them. The problem is that this requires painstaking examination of thousands of images—and so far, only a handful have been confirmed.
That’s about to change. In August, Mutlu-Pakdil received funding from the National Science Foundation to develop an AI algorithm that can automate the hunt, scanning astronomical images for the telltale signs of dwarf galaxies that might be hard to find otherwise. She’s collaborating with researchers at the University of Arizona and the University of Tampa to make these elusive galaxies easier to pinpoint.
“We are at the forefront of the astronomy field here at Dartmouth, and this AI algorithm will help us push the limits in astronomy discoveries even further,” says Mutlu-Pakdil.
The biggest dwarf galaxies are less than one tenth the size of the Milky Way, and the ones Mutlu-Pakdil is searching for are about 100 times smaller still. These tiny galaxies are the building blocks of larger galaxies and have a higher ratio of dark matter to total matter. Studying them helps astrophysicists understand the role of dark matter and the processes by which galaxies evolve. However, their faintness and small size make them difficult to detect, so most of what we know about dwarf galaxies is based on systems in or near the Milky Way.
“We don’t know whether the dwarf galaxies around the Milky Way are representative of all the dwarf galaxies out there, or if they’re very specific to that local environment,” says Mutlu-Pakdil. “Our AI algorithm will help us push the limits and find these tiny dwarf galaxies everywhere, even in isolated areas.”
Because existing algorithms can’t reliably detect dwarf galaxies, and manual searches are exceedingly time-consuming and tedious, Mutlu-Pakdil decided to train AI specifically for the hunt.
“In this big data era, we cannot manually look for things anymore,” says Mutlu-Pakdil. “The hardest part is often identifying the galaxies. Once discovered, they’re more straightforward to study, though follow-up can still be challenging depending on how faint the object is and how distant.”
Mutlu-Pakdil’s team trained the AI algorithm by showing it images of known dwarf galaxies and similar-looking faint objects that might be confused for dwarf galaxies. The researchers will further improve the algorithm by adding each newly discovered dwarf galaxy to its training set.
Using an early version of the AI algorithm, Mutlu-Pakdil’s team has already identified over 100 possible dwarf galaxies that they plan to investigate further. They’ve confirmed that three candidates are indeed dwarf galaxies, including one that Guarini graduate student Guinevere Herron named “Kamino” after a remote planet in Star Wars: Episode II.
“We're planning to get follow-up observations on these systems to really understand how these tiny galaxies form, and what they tell us about how the universe works,” says Mutlu-Pakdil.
When the team spots a potential dwarf galaxy, they launch an extensive campaign to gather additional images and data from multiple telescopes at different wavelengths.
“Each wavelength gives us different clues about a galaxy’s past,” says Mutlu-Pakdil. “From these data, we can estimate the ages of its stars, determine whether it still contains gas that could form new stars, and reconstruct its star-formation history. Because most dwarf galaxies formed early, they offer valuable information about the conditions in the early universe.”
In parallel with their research, Mutlu-Pakdil’s team will develop a series of videos that describe their process of scientific discovery. The videos, designed for middle schoolers and beyond, will be shared in 2026 on SciAll’s YouTube channel.
“Science is super fun and I want everyone to enjoy it,” says Mutlu-Pakdil. “By sharing our process and discoveries, we hope to foster an appreciation for science, and inspire young people to pursue careers in STEM.”