报告题目：Efficient Simulation Optimization under Uncertainty via Multi-fidelity Modeling and Analysis
报告人：Dr.Jie XU(Assistant Professor of Operations Research at George Mason University, Fairfax, VA，U.S.A.)
Simulation optimization provides a powerful and general tool to optimize the design of complex systems. However, major difficulties arise when the simulation is time-consuming and computational budget is limited. In this talk, we propose a new multi-fidelity framework that utilizes information from low-fidelity models to improve the computational efficiency of simulation optimization. Through a new approach known as ordinal transformation, we demonstrate the benefit of the multi-fidelity framework and design an optimal sampling allocation rule to efficiently use the limited computation budget to search for an optimal solution. Preliminary experiments demonstrate that the new ordinal transformation framework can lead to significant computational savings.
Dr. Jie Xu is an Assistant Professor of Operations Research at George Mason University, Fairfax, VA, USA. He received his Ph.D. degree in Industrial Engineering and Management Sciences from Northwestern University, Evanston, IL, USA, in 2009. His research interests are the modeling, simulation, and optimization of complex stochastic systems. His expertise includes simulation optimization, data analytics, rare event simulation, computational intelligence, and their applications in revenue management, production planning, health care, risk management, and energy systems. His work has been sponsored by the National Science Foundation, Air Force of Scientific Research, Oak Ridge Associated Universities, and the Office of Naval Research.