喻园管理论坛2019年第77期(总第528期)
演讲主题: Dynamic Data Driven Simulation and Data Driven Simulation Modeling: Toward human directed autonomy for wildfire monitoring using multiple UASs.
主 讲 人: 胡晓琳, 美国佐治亚州立大学教授
主 持 人: 胡 斌，管理科学与信息管理系教授
活动时间: 2019年6月19日（周三）9:00-11:00
活动地点: 学院119室
主讲人简介:
Dr. Xiaolin Hu is a Professor in the Computer Science Department of Georgia State University. He received his Ph.D. degree from the University of Arizona, M.S. degree from Chinese Academy of Sciences, and B.S. degree from Beijing Institute of Technology in 2004, 1999, and 1996 respectively. His research interests include modeling and simulation theory and application, complex systems science, agent and multi-agent systems, and advanced computing in parallel and cloud environments. He has applied modeling and simulation to a wide range of applications, covering computer science and interdisciplinary research areas such as ecological science, behavioral and social science, and public health modeling and simulation. He served as program chairs/ coordinators for many international conferences/symposiums in the field of modeling and simulation, and is associate editors for several modeling and simulation journals, including ACM Transactions of Modeling and Computer Simulation (TOMACS), and Simulation: Transaction of The Society for Modeling and Simulation International. Dr. Hu received the National Science Foundation (NSF) CAREER Award in 2009.
活动简介:
Computer modeling and simulation deals with constructing simulation models to capture the dynamic behavior of complex systems and executing the simulation models to analyze and predict system behavior. Traditionally, data has been mainly used to help model calibration (e.g., parameter calibration) and simulation validation (e.g., comparing simulation results with historical data). As the quality and availability of data increases dynamically in recent years, there is increasing interests in using data in computer modeling and simulation in new ways. In this talk, I will present our work on dynamic data driven simulation and data driven simulation modeling. In dynamic data driven simulation, we developed a data assimilation framework based on Sequential Monte Carlo (SMC) methods that allows real time sensor data to be continuously assimilated into running simulation models to support real time prediction and analysis for complex systems. This contrasts with traditional simulations where simulations are decoupled from real time data and thus are mainly used as offline tools. In data driven simulation modeling, we developed a framework for discovering simulation models in an automated way for mobile agent-based applications (e.g., flocks of birds, schools of fish, pedestrian crowds, and road traffic). The developed framework makes it possible to utilize behavior data observed from a system to discover candidate simulation models, which, when simulated, produce behaviors similar to what have been observed. The principles of dynamic data driven simulation and data driven simulation modeling and the developed frameworks will be presented, and experiment results will be shown to demonstrate the utilities of the developed data driven approaches.
