| 讲座题目 | Statistical Analysis of Trajectories on Manifolds | ||||
|---|---|---|---|---|---|
| 主办单位 | 信息科学与工程学院 | ||||
| 联合主办单位 | |||||
| 讲 座 人 | 苏敬勇 | 讲 座人 职 称 |
副高 | 主持人 | 潘正祥 |
| 讲座类型 | 自然科学 | 讲座对象 | 全校师生 | 时 间 | 2019-09-11 10:30 |
| 地 点 | 信息学院C1-206会议室 | ||||
| 讲 座 人 简 介 |
Dr. Jingyong Su is an associate professor in the department of Mathematics & Statistics at Texas Tech University, USA. He completed a Ph.D. in statistics at the Florida State University in 2013. Previously, he received both B.S. and M.S. degrees in electrical engineering from Harbin Institute of Technology in 2006 and 2008 respectively. His current research interests include statistics on manifolds, pattern recognition, computer vision and medical imaging. Dr. Su has published multiple papers on top-rated journals and conferences, such as IEEE TPAMI, IJCV, Annals of Applied Statistics, JIVC, CSDA, CVPR and ICPR etc. | ||||
| 讲 座 主要内容 |
In this research we proposed a comprehensive framework for registration and analysis of manifold-valued processes. Functional data analysis in Euclidean spaces has been explored extensively in literature. But we study a dierent problem in the sense that functions to be studied take values on nonlinear manifolds, rather than in vector spaces. Manifold-valued data appear frequently in shape and image analysis, computer vision, biomechanics and many others. The non-linearity of the manifolds requires development of new methodologies suitable for analysis of manifold-valued data. We propose a comprehensive framework for joint registration and analysis of multiple manifold-valued processes. The goals are to take temporal variability into account, derive a rate-invariant metric and generate statistical summaries (sample mean, covariance etc.), which can be further used for registering and modeling multiple trajectories. | ||||
欢迎关注
