Computational tools for society’s most complex challenges

Cathy Wu, an associate professor in the MIT Department of Civil and Environmental Engineering and the Institute for Data, Systems, and Society, and a principal investigator in the Laboratory for Information and Decision Systems, researches machine learning and reinforcement learning (RL) to improve complex systems including transportation.

In 2018, during her final year of a PhD at the University of California at Berkeley, Wu applied RL to a traffic problem: automatically analyzing the potential traffic flow impact of autonomous vehicles across different traffic networks. Wu said the research went viral. After a postdoc at Microsoft working on RL theory, she returned to MIT as faculty. Over the following two years, her further attempts to apply RL to traffic problems failed. "That was stressful," Wu said, adding it was unclear whether the problem was her, her students, the traffic domain, or RL itself.

In 2022, Wu and her students identified that RL algorithms are so sensitive that an algorithm that works on one problem may not work on a closely related one. In 2023, the team devised a way around that sensitivity: RL may not train well on 90 percent of a group of problems, but it can train quite well on 10 percent. By training models on problems that solve and generalize well, the resulting models perform well collectively on related problems, including those not solved through direct training. The researchers designed an algorithm to determine which problems to use RL to train, improving training efficiency by up to 30 times, so what normally would require 100 training models may need only three.

The work "gave me back the confidence that reinforcement learning can play an important role in solving hard optimization problems, including in transportation," Wu said. A good chunk of her group now works on contextual RL.

More recent research applies RL to transportation policy: eco-driving measures in which vehicle speeds are intelligently controlled to reduce excessive stopping and starting could reduce vehicle emissions by between 11 and 22 percent. Wu called it "a demonstration that RL can be used to inform transportation policy on problems of practical importance."

Wu has received a 2023 National Science Foundation Faculty Early Career Development Award and the Ole Madsen Mentoring Award in 2025. She advises students facing complex problems to "be patient. Start small."