统计与数据科学系系列学术报告之三百八十二

 

报告时间:9月27日(周二)14:00-15:30

报告地点:腾讯会议号:604-991-625,密码:678998

主  持 人:郁文  教授

报告题目:Semi-Distance Correlation and Its Applications

报  告 人:钟威  教授   

                厦门大学   王亚南经济研究院、经济学院    统计学与数据科学系

个人简介:钟威教授现任厦门大学王亚南经济研究院、经济学院统计学与数据科学系系主任,博士生导师。2012年获得美国宾夕法尼亚州立大学统计学博士学位,2014年和2017年分别破格晋升副教授和教授,2018年入选厦门大学“南强青年拔尖人才”(A类),国家自然科学基金优秀青年基金获得者(2019),福建省杰出青年基金获得者(2019)。主要从事高维数据统计分析、统计学习算法、计量经济学等研究,在AOS, JASA, Biometrika, JOE, JBES, Biometrics, AOAS, Statistica Sinica,《中国科学数学》,《数学学报》等国内外统计学权威期刊发表30余篇论文。2016年获得厦门大学第五届英语教学比赛一等奖,2020年获得厦门大学第十五届青年教师技能比赛特等奖,2021年获得厦门大学教学创新大赛一等奖,2021年获得福建省“向上向善好青年”称号,2022年获得教育部霍英东高等院校“青年科学奖”二等奖。

摘      要:We propose a new measure of dependence between a categorical random variable and an arbitrary-dimensional continuous random vector, named semi-distance correlation. Our dependence measure equals zero if and only if they are independent, and takes values between zero and one after standardization, just like a correlation coefficient. Two important applications of semi-distance correlation are considered. One is the test of independence between a categorical random variable and an arbitrary-dimensional continuous random vector. The asymptotic distributions of the test statistic are established. The asymptotical normal null distribution allows use to compute critical values and p-values efficiently when the dimension of continuous variables approaches the infinity. The second is a groupwise feature screening procedure for classification problems based on the semi-distance correlation. Its sure screening property is proved. Both Monte Carlo simulation studies and a real-data application are presented to demonstrate the excellent finite sample performances of the proposed procedures. 

 

统计与数据科学系

2022-9-13