时间：2018年4月4日 下午2:30-3:30
地点: 明德主楼1016会议室
题目：High Dimensional Sliced Inverse Regression
摘要：In this talk, we will briefly review some recent theoretical advances in sliced inverse regression from the statistical decision theory perspective. We then compare them with the existing results in linear regression and propose a new program which aimed at transplanting the rich theories and methodologies for linear regression into sliced inverse regression or sufficient dimension reduction. We will present several supportive evidences for this speculation including the Lasso-SIR algorithm and the optimal detection boundary of single index models.
简介：林乾，清华大学统计中心助理教授。2010在麻省理工数学系获得博士学位。2014-2017在哈佛大学统计系做博士后研究。2017年8月至今在清华大学任教。从事高维充分性降维，统计计算，及统计机器学习等方面的研究。