Mengdi Wang

Assistant Professor

Department of Operations Research and Financial Engineering
Princeton University
Sherrerd Hall 226, Princeton, NJ 08544
Email: mengdiw at princeton dot edu

Mengdi Wang is interested in data-driven stochastic optimization and applications in machine and reinforcement learning. She received her PhD in Electrical Engineering and Computer Science from Massachusetts Institute of Technology in 2013. At MIT, Mengdi was affiliated with the Laboratory for Information and Decision Systems and was advised by Dimitri P. Bertsekas. Mengdi became an assistant professor at Princeton in 2014. She received the Young Researcher Prize in Continuous Optimization of the Mathematical Optimization Society in 2016 (awarded once every three years), the Princeton SEAS Innovation Award in 2016, and the NSF Career Award in 2017.

I have two openings for PhD students and two openings for postdocs. Feel free to email me if interested.

[my cv]

Stochastic Methods for Data-Driven Optimization

Data-driven optimization arises from machine learning, data analysis, empirical operations research.

  • Scalable algorithms for large-scale offline data
  • Online optimization algorithms processing data streams
  • Fundamental complexity and uncertainty quantification of data-driven problem
  • Applications in risk management and machine learning
Highlight: Stochastic Nested Composition Optimization , a rich new problem class motivated from risk management, data analysis and system control.

[learn more]

Nonconvex Statistical Optimization

Nonconvex optimization permeates data analysis ranging from sparse learning to principal component analysis.
  • Develop computable approximate solutions with strong theoretical guarantees.
  • Decomposition of high-dimensional nonconvex optimization with special structures
  • Fundamental complexity of nonconvex optimization.
  • Applications in dimension reduction of high-dimensional data.
[learn more]

Dynamic Programming and Reinforcement Learning

A mathematical programming approach towards reinforcement learning.
  • Stochastic methods for online solution of unknown systems
  • Sample complexity of black-box learning
  • Optimization modeling, stochastic methods, and dimension reduction.
  • Application in modeling of clinical data.
[learn more]