New directions in optimization

Thursday, February 25, 2021

“I started out as a student in mathematics,” said Professor Yuying Li. “It was a huge, scary move to come to Waterloo for computer science. Thankfully, it has worked out.” 

After completing an undergraduate degree in mathematics in her native China, Li moved to Waterloo in 1983 to study artificial intelligence (AI). In the course of her masters and doctoral degrees in computer science, she changed direction when she found herself increasingly drawn to optimization. After 16 years of research at Cornell University, Li returned to her alma mater as a faculty member. “My research lies at the intersection of computer science and math, so the opportunity to join a faculty that houses both under the same umbrella was appealing to me,” she remembered. “Being part of the Faculty of Mathematics has ultimately allowed to me pursue exciting interdisciplinary work.” 

For the past 15 years, Li’s research has centered on computational optimization to solve complex, real-world problems. She designs, analyzes, and implements efficient algorithms for computing problems related to finance, primarily. In an ironic twist of fate, her focus on optimization has brought her full circle as she relies on AI and machine learning applications in her everyday work. 

Recently, Li has teamed up with Professor Peter Forsyth, a colleague in the Cheriton School of Computer Science, to develop optimal strategies for allocating retirement savings through leveraging machine learning applications. “Essentially, we mine historical market data and use a machine learning approach to model the problem,” Li explained. “Then we use computation to discover optimal allocation decisions.” 

Within most wealth management companies, allocation is a simply function of time, age, and amount of savings accumulated. Li’s research, in contrast, adds layers of complexity through detailed market analysis. In recognition of the value of her work, a wealth management firm is providing financial support for students to assist with research. 

Compared to most data mining problems, financial problems present a particular challenge. “The amount of data we’re working with is typically very small,” shared Li. “It requires creative thinking to take a data-driven approach to solve a problem based on so little information.” 

For the past two years, Li served as the first director of the graduate program in data science. She stepped down from the role at the end of 2020, but she’s confident that the program will continue to give students a competitive edge. “The timing is right,” she expressed. “I always believed our data graduate program had potential, but I’ve been extremely pleased by the tremendous interest it has received from students and industry leaders alike.” 

Though she first wanted to be a mathematician, Li’s fascination with the synergy between mathematics and computer science has catalyzed solutions to pressing real-world problems. “For me, career satisfaction comes from connecting my research to the outside world,” she shared. “Fortunately, Waterloo is the right place to be.”