数苑学术沙龙第五十五讲 Sufficient Dimension Reduction: A generalization of Canonical Correlation Analysis

发布时间：2018-05-17 作者：浏览次数：

Speaker:	於州教授	DateTime:	2018年5月24日（周四）下午2点30分—3点30分
Brief Introduction to Speaker: 於州教授，华东师范大学（上海东方学者）		Place:	6401
Abstract:It is increasingly interesting to model the relationship between high dimensional covariates and the corresponding response. Sliced inverse regression (SIR) is an innovative and effective method for sufficient dimension reduction and data visualization, we in this article formulate the Sparse Sliced Inverse Regression (Sparse SIR) as a generalization of the CCA approach, and propose a new sparse SIR method that recasts high-dimensional SIR as a convex problem. Theoretical results show that Sparse SIR can consistently estimate the true dimension reduction space with an overwhelming probability in ultra-high dimensions. Numerical results also demonstrate the competitive performance of our proposed estimation methods