CORMSIS Seminar - "Empirical dynamic programming and random operators", Talk by William Haskell (University of Singapore) Seminar
- Time:
- 14:00 - 16:00
- Date:
- 18 December 2018
- Venue:
- Building 04, Room 4055, University of Southampton, Highfield Campus, SO17 1BJ
For more information regarding this seminar, please email Huifu Xu at H.Xu@southampton.ac.uk .
Event details
In this talk, we consider simulation-based algorithms for large-scale dynamic programming. We develop a suite of tools that we call 'empirical dynamic programming' (EDP). The EDP algorithms are easily implementable, and naturally support function approximation for large-scale implementation. We demonstrate the favorable performance of these algorithms. Our convergence analysis for EDP is based on iteration of random operators. As one of our key contributions, we develop a new method for convergence analysis of iteration of random operators, and show that it has broad applicability to many algorithms.
Speaker information
William B. Haskell, University of Singapore. William completed his Ph.D. in Operations Research at the University of California Berkeley in 2011 with J. George Shanthikumar and Z. Max Shen. Then, he spent three years as a visiting assistant professor at the University of Southern California before joining the National University of Singapore in 2014. His current research interests cover risk-aware decision-making, data-driven optimization, and scalable algorithms.