应beat365官方网站的邀请，中国人民大学朱利平教授近期将访问我院，来访期间将为我院师生做学术报告：

        报告时间：5月27日下午16：40—17：40
        报告地点：理科楼202
        报告题目：Cumulative Divergence With Application to Forward Regression

        摘要：Feature screening plays a pivotal role in analysis of ultra-high dimensional data. Various screening methods are proposed based on marginal correlations  between a response variable  and a covariate vector. Despite their encouraging applications, the existing methods often suffer from model mis-specification and  presence of outliers. In addition, an appropriate screening size needs to be carefully specified for existing methods. This paper aims to develop a model-free forward screening procedure based on cumulative divergence (CD), a new correlation metric proposed in this work. The CD is a robust measure to test regressional independence of a covariate  in the presence of outliers or extreme values. This robust property makes CD-based screening method very appealing for handling high dimensional noisy data. The newly proposed forward CD-based screening procedure accounts for joint effects among features. Moreover, it can automatically determine the number of features that are to be screened out. Under some regularity conditions, we show that the proposed method is screening-consistent. Our simulation study shows the CD-based feature screening procedure is quite promising. We further illustrate the proposed method through a real data example.

        报告人简介：朱立平博士为中国人民大学统计与大数据研究院教授，2006年于华东师范大学获得博士学位，同年任华东师范大学助理教授。2013年入选教育部新世纪优秀人才计划，2015年获得国家自然科学基金委“优秀青年基金”等资助。在统计顶级刊物Annals of Statistics，Journal of the Royal Statistical Society Series B，Journal of the American Statistical Association，Biometrika等顶级杂志上发表超过15篇文章。他的主要研究兴趣有半参数建模、高维数据分析、充分降维、变量选择等领域。 

        欢迎感兴趣的师生参加！