Data-driven State-Space Model for Harsh Scenarios
报告人：Feng Yin
报告人单位：The Chinese University of Hong Kong, Shenzhen, China
时间：22年6月21日（周二）晚上 7：30~8：30
线上腾讯会议号：177-701-315 密码：610064
摘要：Studies on time-variant complex systems were awarded the 2021 Nobel prize in Physics and become the spotlight nowadays. To model time-variant complex systems often requires the use of state-space models, which have been widely used in various sectors, from military to civilian applications, including missile interception, autonomous navigation, epidemic tracking, and so on. Classical state-space models have to assume a pair of known and fixed state transition function and observation function, based on which the desired system states are eventually estimated. However, when facing complex scenarios, we often cannot select a proper pair of candidates from the known transition functions and observations functions, or they are barely expressible. Mismatch of the underlying functions will incur significant performance degradation. Adopting machine learning models and data-driven methods is a promising way to address the above problems. In this tutorial-type seminar, I will start from the classical state-space model-based Kalman filter (KF), Extended KF filter, particle filter and walk rhythmically to various recent data-driven state-space models, including the deep SSM and Bayesian GPSSM with enhanced modeling capacity. Some recent results and emerging use cases will be presented in the end.
报告人简介：Feng Yin received his B.Sc. degree from Shanghai Jiao Tong University, China, and his M.Sc. and Ph.D. degrees from Technische Universitaet Darmstadt, Germany. From 2014 to 2016, he was a postdoc researcher with Ericsson Research, Linkoping, Sweden. Since 2016, he has been with The Chinese University of Hong Kong, Shenzhen, as an assistant professor. His research interests include statistical signal processing, Bayesian learning and optimization, and sensory data fusion. He has published more than 30 journal papers, 40 conferences, and 20 patents/standards. He was a recipient of the Chinese Government Award for Outstanding Self-Financed Students Abroad in 2013 and the Marie Curie Scholarship from the European Union in 2014. He was the finalist for the IEEE CAMSAP best paper award in 2013. He is an IEEE senior member and currently serving as the Associate Editor for the Elsevier Signal Processing Journal.
邀请人：宋恩彬 教授
