4/10/25 | 4:15pm | E51-376
Cathy Wu
Thomas D. and Virginia W. Cabot Career Development Associate Professor
MIT
Abstract: The increasing complexity of modern mobility systems is challenging traditional engineering practices. A fundamental challenge is that different engineering requirements lead to distinct optimization and control problems, each necessitating specialized techniques that can take years to develop. As a result, researchers are increasingly turning to machine learning (ML) to automate the development of solvers. However, recent findings—including our own—indicate that ML methods can be brittle and often underperform compared to classical solvers. To address this, we propose a learning-guided optimization approach that combines the strengths of ML with established optimization techniques. The focus is on creating algorithms that can automatically customize optimization methods based on the specific characteristics of different problem instances. We have developed learning-guided optimization algorithms for branch-and-cut methods, rolling horizon optimization, and large neighborhood search. Experimentally, these algorithms demonstrate the potential to accelerate state-of-the-art optimization solvers by a factor of 2 to 7. We will discuss applications in areas such as mixed-integer linear programming, flexible job shop scheduling, vehicle routing, and multi-agent pathfinding.
Bio: Cathy Wu is an Associate Professor at MIT in LIDS, CEE, and IDSS. She holds a Ph.D. from UC Berkeley, and B.S. and M.Eng. from MIT, all in EECS, and completed a Postdoc at Microsoft Research. Her research advances machine learning for control and optimization in mobility. She is broadly interested in AI for Engineering. Cathy has received a number of awards, including the NSF CAREER, PhD dissertation awards, and publications with distinction. She serves on the Board of Governors for the IEEE ITSS, is a Program Co-chair for RLC 2025, and is an Associate Editor (or equivalent) for ICML, NeurIPS, and ICRA. She is also spearheading efforts towards reproducible research in transportation, including co-founding the RERITE Working Group.