报告题目:Sure ExplainedVariability and Independence Screening

报 告 人：陈敏 研究员 中科院

报告时间：2019年5月21日14:30-16:00

报告地点：数学楼一楼第二报告厅

报告摘要：

In the era of Big Data, extracting the mostimportant exploratory variables available in ultrahigh dimensional data plays akey role in scientific researches. Existing researches have been mainlyfocusing on applying the extracted exploratory variables to describe thecentral tendency of their related response variables. For a response variable,its variability characteristic is as much important as the central tendency instatistical inference. This paper focuses on the variability and proposes a newmodel-free feature screening approach: sure explained variability andindependence screening (SEVIS). The core of SEVIS is to take the advantage ofrecently proposed asymmetric and nonlinear generalized measures of correlationin the screening. Under some mild conditions, the paper shows that SEVIS notonly possesses desired sure screening property and ranking consistencyproperty, but also is a computational convenient variable selection method todeal with ultrahigh-dimensional data sets with more features than observations.The superior performance of SEVIS, compared with existing model-free methods,is illustrated in extensive simulations. A real example inultrahigh-dimensional variable selection demonstrates that the variablesselected by SEVIS better explain not only the response variables, but also thevariables selected by other methods.

报告人简介：

陈敏研究员现担任中国科学院政府行政管理系统分析研究中心主任、全国统计方法应用技术标准化委员会主任委员、《数学与统计管理》主编、中国数学学会副理事长、中国统计教育学会副会长等职，研究方向为金融统计理论与方法、非线性时间序列的统计分析、非参数统计估计和检验的大样本理论、生物统计的理论和方法、应用统计、大数据分析与处理的统计理论和算法研究。