Machine learning research requires the ability to juggle a lot of topics in an ever-shifting field. New Assistant Professor Jake Abernethy has excelled in it because his first career was comedy juggling.

Anyone can learn to juggle after a few months, but keeping the balls in the air while making sure the jokes land takes skill.

“When you’re doing a show, you have these props, an objective, that you’re weaving the story around and telling jokes around this existing narrative,” he says.

Machine learning, which many believe will drive the future of computing, if not society, presents an equal challenge. “Machine learning requires knowledge of statistics and probability, linear algebra, coding, optimization — skill sets that are quite rare,” Abernethy says. “But future companies are going to need this skill set from more and more of their workforce. It’s exciting to be on the cusp of this.”

Building Skills

Abernethy has always obsessively approached new skills. In high school, it was juggling. His step-mother gave him his first set of juggling equipment when he was 14. “While most kids were doing their homework, I was on the front lawn juggling,” he says. “Within a year, I had mastered the unicycle, was juggling five balls, and had picked up enough tricks to do it professionally.”

Although he found success in juggling, he became increasingly disillusioned with his high school in Amherst, Massachusetts. He eventually dropped out at the age of 15, but soon thereafter began taking classes at the Harvard Extension School and earned his GED at 16. He enrolled at the University of Massachusetts Amherst, where both his academic and entertainment careers began to flourish. He won a university-wide comedy competition and even opened for Dave Chappelle and Sinbad. Around this time, he applied to transfer to MIT, where he finished in 2002 with a B.S. in mathematics.

Despite his whirlwind academic experience, by senior year Abernethy had burnt out on his math degree. Rather than corporate jobs or graduate school, he chose to ride a bicycle across the country with a few friends before figuring out his next move.

“I was ambitious with learning skills. I’d done juggling, done math, but I needed something else, something new,” he says.

Pure math was intellectually engaging but didn’t have enough practical application for Abernethy, so a professor suggested he look into the burgeoning field of bioinformatics. Not only did Abernethy pursue it, he also added in another new skill — learning French. He applied for an internship at Insead, a French business school, to combine the two.

Those four months in Fontainebleau, France, in 2004 introduced Abernethy to the ultimate skill to master: machine learning. At that time, the field was small and hardly considered central to computer science, but the mix of theory and application appealed to Abernethy. “You get excited about something and then you’re obsessed with it,” he says. “Machine learning is a good thing to get obsessed with because it keeps growing.”

Abernethy’s obsession led him to earn a master’s in computer science at the Toyota Technological Institute in Chicago, where SCS Chair Lance Fortnow was one of his professors. He then went on to earn his Ph.D. in CS at University of California, Berkeley, studying with Professor Peter Bartlett. Following his doctorate, Abernethy was a fellow at the Simons Foundation under renowned computer scientist Michael Kearns. Eventually, he was hired as an assistant professor at the Department of Electrical Engineering and Computer Science at the University of Michigan, where he found a way machine learning could benefit everyone.

Using Machine Learning to Help Flint

The highlight of Abernethy’s four years at Michigan was using machine learning to help mitigate the Flint water crisis. When Flint, Michigan, changed the water source to the Flint River, the water was so corrosive it caused lead in old pipes to leak into the drinking water supply.

One of the major delays in fixing this problem was figuring out which homes were  prone to contaminated water and where the lead pipes were in the city. Flint did not maintain accurate records of its infrastructure, complicating the issue further.

Yet Google became interested in supporting efforts in the city, providing $150,000 in funding to Abernethy and a team at UM Flint for a website and app for residents to learn if their water might be contaminated. Abernethy and his student team used machine learning to determine which homes might have dangerous water and which pipes might be compromised. The predictive model was based on available city data: water test results, records of the service lines that deliver water to homes, information on parcels of land, and water usage. Despite joining Tech, Abernethy continues to engage with Flint’s policymakers on the project.

Abernethy is still passionate about this research and continues to author papers on the topic, but he hasn’t given up on socially minded research since moving to Tech. “I realized Atlanta would have these opportunities, so I’m wondering what my next socially focused project will be,” he says.

Teaching the Next

This community forward research approach isn’t the only thing he’s bringing with him to Tech.

As a professor, Abernethy prides himself on being down to earth. “Having been a professional juggler for some part of my life, I probably like the showmanship aspect of teaching more than I should,” he says. “I like making it a fun experience and making myself approachable. I want to feel that the students and I are working together.”

At his previous position, he launched a student machine learning initiative called the Michigan Data Science Team. Together, the team worked on machine learning challenges and eventually collaborated on the Flint project. Abernethy himself mostly offered advice, funding, and constructive questions, but let his team organize themselves. It was a very effective model he hopes to bring here.

Machine learning is a field he wants his students to get just as excited about as he is. After all, Abernethy is a high school drop-out former comedic juggler, and now he’s one of the leading researchers in this area.