2016哈工大第706期光熙论坛预告

报告题目：High-Dimensional Particle Filtering and Multi-sensor Multi-target Tracking 

报告地点：哈尔滨工业大学综合楼701

报告时间：2016年6月1日（周三）上午9:30至11:30

报告人：Mark John Coates, IEEE Senior Member, Associate Professor, Department of Electrical and Computer Engineering, McGill University.

报告摘要：

Sequential Monte Carlo methods (or particle filters) can be used to track the posterior distribution of a state by updating a set of weighted particles when new data become available. When the state dimension is high or observations are highly informative, many algorithms suffer from the particle degeneracy phenomenon: almost all particles have extremely small weights after a few timesteps. This particle degeneracy phenomenon, associated with the curse of dimensionality, leads to poor representation of the posterior and inaccurate state estimates. Particle flow algorithms avoid the importance sampling step in particle filters; they migrate particles from the prior distribution to the posterior distribution, via a “flow” that is specified through a partial differential equation. In this talk, I will describe our recent research towards combining the particle flow and particle filtering methodologies, leading to a method that has the impressive performance of the particle flow algorithms but also the theoretical performance guarantees of the particle filter. I will also discuss our recently developed strategies for tracking multiple targets using sensor networks. We have developed novel algorithms for combining the measurements from multiple sensors in a computationally tractable fashion.

报告人简介：

Mark John Coates, IEEE Senior Member, 加拿大麦吉尔大学电子与计算机工程专业副教授；1995年于澳大利亚阿德莱德大学获得计算机系统工程学士学位，1999年于英国剑桥大学获得信息工程博士学位；1999-2001，美国莱斯大学，博士后、讲师；2002年起在麦吉尔大学工作；2012-2013，英国Winton Capital Management作为Senior Scientist从事研究；2007-2011，任IEEE Transactions on Signal Processing期刊Associate Editor；2012-2015任IEEE Signal Processing Letters期刊Senior Area Editor；2006年，研究团队获得NSERC协同创新奖；已发表30余篇期刊论文与80余篇会议论文，他引量超过4800余次。作为负责人以及共同负责人承担超过20项基金以及合约项目；

作为期刊/国际会议审稿人，已审阅90多篇期刊论文、150多篇会议论文。包括IEEE Trans. Signal Processing, IEEE Signal Processing Magazine, IEEE Trans. Sensor Networks, IEEE Trans. Image Processing, IEEE Trans. Information Theory, IEEE Trans. Automatic Control, IEEE Trans. Vehicular Technology, IEEE/ACM Trans. Networking, ACM Trans. Internet Technology, Computer Communications, Eur. J. Advances in Signal Processing, IEEE/ACM Int. Symp. on Inf. Proc. in Sensor Networks, IEEE Infocom, Neural Information and Processing Systems (NIPS). 作为基金项目计划评审人员，已评审20多项基金计划，相关基金组织包括NSERC Collaborative Resarch and Development Projects, NSERC Strategic Projects, NSERC Discovery Grants, National Science Foundation Major Grant, National Science Foundation Panel Review, National Science Foundation Career Award.

研究领域包括：通信与传感器网络，统计信号处理，以及贝叶斯与蒙特卡罗方法研究。

个人主页：http://www.ece.mcgill.ca/~mcoate/