报告时间：12月14日周四上午9：30-10：30
报告地点：法商南楼228会议室
报告人：林乾
报告题目：On consistency of sliced inverse regression
报告摘要：The recent advances of the theoretical developments of sliced inverse regression depend on a rather obscure condition, the sliced stability inequality. In this talk, we will briefly discuss the motivation of the sliced stability inequality and show that this inequality under the existence of $2+\epsilon$ moment of the predictors. As a direct corollary, we prove the consistency of sliced inverse regression under the weakest condition in the literature.
报告人简介：林乾博士在麻省理工学院获得博士学位，随后在哈佛大学统计系从事博士后研究，现任职于清华大学统计科学中心。研究方向为随机矩阵理论，高维数据降维，曾在Annals of Statistics等顶级统计学刊物上发表文章数篇。