[Mingli Lecture Hall, Issue 44,2020] 12-15 Assistant Professor Peng Yijie, Peking University: gradient-based stochastic simulation optimization
Reporter: Assistant Professor Peng Yijie, Peking University
Time: 11:00:00-12:00,15 December 2020
Location: Main Building 418
Introduction to content:
Actual risk measures of random systems do not usually have an analytical form, but can be estimated through simulations.Theclassical simulation optimization problem studies the maximization of performance expectations of stochastic system samples, and the gradient-based algorithm is the most efficient way to optimize high-dimensional complex stochastic simulation systems.Stochastic gradient estimation is the central problem in which the reporter will review the classical approach.How to obtain unbiased stochastic gradient estimates of the performance of discontinuous samples with structured parameters is a core theoretical problem in the simulation field for decades, the reporter proposes a generalized likelihood ratio method to systematically solve this problem and gives its application to financial, artificial intelligence, etc.Risk value (VaR) is a commonly used financial risk measure, and the optimization of risk value has an important significance in financial risk management and portfolio optimization.The reporter will give a multi-scale stochastic gradient descent algorithm based on the sensitivity estimation of the generalized likelihood ratio, which is shown to converge to the optimal solution of the risk value with probability 1, and illustrate the effectiveness of the proposed method and the application prospects in practical problems.
Speaker profile:
Peng Yijie, Assistant professor and doctoral supervisor of Guanghua School of Management, Peking University.He graduated with his bachelor's degree from the School of Mathematics and Statistics of Wuhan University and received his PhD from the School of Management of Fudan University.He was a postdoctoral fellow and an assistant professor at the University of Maryland and George Mason University, respectively.The main research directions include simulation modeling and optimization, financial engineering and risk management, artificial intelligence, health care, etc.He has presided over a number of scientific research fund projects, including National Excellent Youth Science Foundation Project, National Youth Science Foundation Project, Beijing Youth Key Individual Project, etc.He has published over 20 academic papers in high-quality journals such as《Operations Research》,《INFORMS Journal on Computing》and《IEEE Transactions on Automatic Control》.He has won the 2019 INFORMS Outstanding Simulation Publication Award,2017 year IEEE Robotics and Automatic Society Best Paper Award Finalists.At present, he serves as the editorial board of Asia-Pacific Journal of Operational Research Journal and IEEE Control Systems Society Conference, executive director of Financial Engineering and Financial Risk Management Branch of China Transportation Research Association, and member of the Artificial Society Special Committee of China Simulation Association.
(Organized by: Department of Management Science and Logistics, Scientific Research and Academic Exchange Center)
