Everything Qiaomin Xie had heard about the University of Wisconsin-Madison led her to believe it would be an ideal landing spot for the next phase of her academic career.
As an incoming assistant professor in the Department of Industrial and Systems Engineering, she would be joining a faculty whose expertise in areas like optimization, machine learning, and healthcare and manufacturing systems would dovetail with her research in applied probability, stochastic networks and reinforcement learning. The growing data science influence and community on campus would allow for myriad collaborations across departments.
And both future and former colleagues had told her plenty of positives about Madison as a city.
Fortunately, when Xie finally visited in July 2021—about a month after she had accepted her new faculty position in ISyE—Madison lived up to her expectations.
“I love Madison,” she says. “It’s such a wonderful city. It’s perfect in terms of size, in terms of location.”
Xie joins ISyE after spending two and a half years as a visiting assistant professor at Cornell University. She earned her PhD in electrical and computer engineering from the University of Illinois in 2016, spending one semester as a research fellow at the Simons Institute for the Theory of Computing in Berkeley, California, and then was a postdoctoral researcher at Massachusetts Institute of Technology for two years.
Xie’s work on the underlying decision-making policies that drive systems with elements of uncertainty and randomness (known as “stochastic” in mathematics) is crucial to computer networks and systems, such as data centers consisting of thousands of servers to host many applications.
While those applications stem from her training in electrical and computer engineering, Xie notes that the methodologies are natural fits for optimizing performance in systems such as the ride-sharing industry and hospitals—the kind of operations research that’s a hallmark of ISyE—as well as for revenue management.
“The main goal of my research is to understand the fundamental properties of these stochastic systems and to try to build a theoretical foundation of efficient algorithms and also practically useful algorithms that can be implemented in real-world systems,” she says.
More recently, Xie has focused on a newer approach to complement her model-based techniques for algorithm design and performance analysis of various systems. This data-driven decision-making framework leans heavily on reinforcement learning, a machine learning paradigm for training intelligent agents to make sequential decisions under uncertainty, while also drawing on tools from game theory, optimization and statistical analysis.
“The goal here is to build the theoretical foundation for data-driven decision-making,” she says.
Artificial intelligence companies like Google DeepMind have used reinforcement learning to achieve remarkable, eye-catching results in quickly mastering games like Go and chess, and the technique holds promise in areas such as robotics and autonomous vehicles. But it hasn’t yet crossed the considerable divide from games to real-world applications, a gap Xie hopes to close through her work.
Despite her focus on the methodological and theoretical side, Xie says it’s crucial to consider practical constraints and requirements when designing algorithms. And she enjoys helping students see the relevancy of modeling and data in their everyday lives. At Cornell, Xie taught a course called Urban Analytics, an introductory-level data analytics class that focused on applications like ride-sharing systems while teaching students how to clean, visualize and analyze data.
She’ll teach ISyE 624: Stochastic Modeling Techniques during the fall 2021 semester, and she hopes to develop courses covering reinforcement learning and game theory in the future.
“It always feels so great to interact with students and make them realize the material and the knowledge they learn in the course actually can have a strong connection to their real life,” she says.
Author: Tom Ziemer