Zooniverse Experiences Quarantine Surge
Zooniverse is the world’s largest and most successful citizen science platform, with over 2 million people contributing to research worldwide. It uses the web to bring together ordinary people and researchers who need help classifying data. Since the COVID-19 quarantine period began, Zooniverse has experienced a huge uptick in use by the public, around five times the usual number of classifications. School of Physics and Astronomy Professor Lucy Fortson says that all the increased activity on Zooniverse research problems means that they are getting completed more quickly than expected.
“It’s a good problem to have,” Fortson says. For example, University Researcher Patrick Wilcox and others just completed Muon Hunter 2—which helps astronomers find elusive muons disguised as gamma rays—ahead of schedule and has just launched Muon Hunter 3 on May 12 rather than sometime this summer as expected. “We’re seeing that happen across many of our 100–150 live projects.”
Fortson is one of the co-founders of the Zooniverse as a platform. The Zooniverse began in 2007 with a single experiment called Galaxy Zoo, which used citizen scientists to sort galaxy images. At the time, Professor Fortson was VP for Research at the Adler Planetarium and had convinced them to start a Citizen Science Division. She met Galaxy Zoo founder, Chris Lintott, and joined forces, with Lintott serving as the first Adler Director of Citizen Science. When Fortson came to Minnesota in 2010, she brought Zooniverse with her, joining the University with the Adler and the University of Oxford as Zooniverse leads. Her work has made the University one of the most active organizations using the platform with over 30 projects, attracting 50,000 participants in the State alone. Fortson recently received the APS Nicholson Medal for Outreach and the College of Science and Engineering Taylor Award for Distinguished Service for her work with Zooniverse.
Zooniverse experiments cover a wide range of subjects from astrophysics to zoology. The latter, projects—watching animals, are particularly popular with families and children during quarantine. “We can’t get out to travel, so it’s like taking a tour of the Serengeti for a few hours,” Fortson says.
Zooniverse projects such as Cedar Creek: Eyes on the Wild, which classifies images of deer, bison, turkey, or wolves in the Minnesota backwoods, are appropriate for children as young as four or five with adult supervision. For teens and undergraduates there are more structured projects, some of which have become popular with teachers looking for science activities that work for distance learning.
There are over 30 University of Minnesota Zooniverse experiments. As an astrophysicist, Fortson is involved directly with Zooniverse experiments such as Muon Hunter which classifies images from the VERITAS telescope array as well as Galaxy Zoo: Clump Scout which aims to create a catalog of galaxies with clumpy star forming regions. Her group also supports other Zooniverse efforts on campus. They also work on machine learning in general.
“Zooniverse is the ideal platform to improve machine learning,” Fortson says. Firstly, the volunteer classifications can be used as training labels. But beyond that, machine learning programs can have biases. Several Zooniverse projects use humans to cross check a certain percentage of the data to validate the machine model and provide new labels to retrain it if needed. In some cases, machines are still not as good as humans at detecting patterns in complex images. Volunteers can detect new and unusual phenomena that can lead to real discoveries such as new classes of galaxies or sightings of rare species in motion-triggered camera trap images (see images below). By combining both human and machine classifiers on Zooniverse, Fortson’s group works to incorporate the strengths of each. “Machine learning is only as good as the data given to it. Zooniverse adds a human in the loop.”