数学与统计及交叉学科前沿论坛------高端学术讲座第115场

报告题目:Partial Consistency and Its Application to Semiparametric Regression Models

报告人:彭衡 教授 香港浸会大学

时间 :2024年9月12日(周四) 14:00~15:00

地点：腾讯会议：752-142-941

参加对象:4556银河国际在线教师及研究生

报告人简介: 彭衡，香港浸会大学数学系教授，博士生导师。2003年从⾹港中⽂⼤学获得统计学博⼠学位，2003年-2006年在普林斯顿⼤学做博⼠后。主要从事⾮参数与半参数模型、模型选择 、 ⾼维数据建模 、 混合模型等领域研究。他是IMS会员，2011-2014担任Statistica Sinica副主编，现为Computational Statistics and Data Analysis副主编。在统计学顶级期刊Annals of Statistics、JASA、Biometrika、Statistica Sinica、TEST 和CSDA等期刊上发表多篇学术论⽂。

报告摘要:For semiparametric regression models, the nonparametric components are often used to reduce the bias of the regression models and improve the stability of the estimate of parametric components in the regression models. Motivated by the partial consistency phenomena which were proposed by Neyman and Scott(1948), regarding the nonparametric components in the model as the so-called incidental parameters and utilizing recent theoretical results in high dimensional statistical modeling, a flexible yet computationally simple approach is proposed to estimate the partially linear models and varying coefficient models, and the variance of errors in the models. The proposed methods are easy to implement and are efficient enough for further inference. Hence, the proposed methods balanced the computation complexity and statistical efficiency of the statistical estimate and provided new insight into the trade-off between the bias and variance of the statistical model estimation.