学术报告第201308期

报告题目：Dynamic  Data Driven Simulation for High Dimensional Spatial-temporal  Systems

报 告 人：  胡晓琳（乔治亚州立大学计算机科学系副教授）

报告时间：  2013年7月10日上午9：00

报告地点：  南一楼中311会议室

邀 请 方：“多谱信息处理技术”国家级重点实验室

Abstract:

Spatial-temporal  systems such as wildfire and road traffic evolve system behavior in both space  and time. Dynamic Data Driven Simulation (DDDS) has shown to be an effective  method to improve simulation accuracy for these systems by assimilating real  time sensor data into a running simulation model. Recently, progresses have been  made in DDDS based on Sequential Monte Carlo methods. Nevertheless, major  research challenges exist in data assimilation for high dimensional  spatial-temporal systems due to the extremely large state space that need to be  covered. These challenges include slow convergence to the true system state,  requirement of a large number of particles, and high computation cost. In this  talk, I will present the background and motivation of our work on DDDS and then  propose several methods to address the challenges of data assimilation for high  dimensional spatial-temporal systems. The methods will be demonstrated using  large-scale wildfire simulations and simulations of crowd behavior in smart  environments. My talk will end with a discussion of the emerging “big data”  trend and how DDDS fits in this new trend by integrating data and simulation  models for improving simulation results.

Bio:
Dr.  Xiaolin Hu is an Associate Professor in the Department of Computer Science at  Georgia State University, Atlanta, Georgia. He received his Ph.D. degree from  the University of Arizona in 2004. His research interests include modeling and  simulation theory and application, agent and multi-agent systems, and complex  systems science. He has served as program chairs for several international  conferences/symposiums in the field of modeling and simulation, and is associate  editors for Simulation: Transaction of The Society for Modeling and Simulation  International, International Journal of Modeling, Simulation, and Scientific  Computing (IJMSSC), and International Journal of Agent Technologies and Systems  (IJATS). Dr. Hu is a USA National Science Foundation (NSF) CAREER Award  recipient.