讲座题目：Variational Bayesian learning for medical imaging data
讲座地点：腾讯会议号：278 316 513
唐年胜，云南大学二级教授、数学与统计学院院长、博士生导师，国家杰出青年科学基金获得者，教育部“长江学者”特聘教授，国家百千万人才工程 暨有突出贡献中青年专家，享受国务院政府特殊津贴，国际数理统计学会会士、国际统计学会推选会员，云南省高等学校教学名师，中国现场统计研究会副理事长，云南省应用统计学会理事长。在 JASA、Annals of Statistics、Biometrika 等刊物发表学术论文 180 余篇，其中 SCI 检索 130 多篇。曾获“霍英东教育基金会第九届高等院校青年教师奖”，省部级科技奖励 9 项。
With the recently developed medical imaging technology, brain images are captured through various scanners. Magnetic resonance image (MRI) and function magnetic resonance image (fMRI) are two widely-used imaging data sources for studying brain disease. In disease diagnosis study, disease prediction based on MRI and fMRI data has received considerable attention over the past years. A key challenging in analyzing MRI and fMRI data is to alleviate the well-known curse of dimensionality. Many Bayesian methods have been developed to address the issue.
This paper aims to introduce variational Bayesian approches to explore the relationship between regions of interest (ROIs) and some specified disease based on high-dimensional generalized linear models, ultrahigh-dimensional generalized tensor regression models, and high-dimensional gaussian graphical models. Some examples associated with MRI and fMRI data analysis are illustrated. Simulation studies demonstrate the empirical validity and improved efficiency of our fused estimators. We illustrate the proposed method with an application.