(一)报告人：蒋滨雁（香港理工大学）

报告题目：Penalized Interaction Estimation for Ultrahigh Dimensional Quadratic Regression

报告摘要：Quadratic regression goes beyond linear model by simultaneously including main effects and interactions between the covariates. The problem of interaction estimation in high dimensional quadratic regression has received extensive attention in the past decade. In this article we introduce a novel method which allows us to estimate the main effects and interactions separately. Unlike existing methods for ultrahigh dimensional quadratic regressions, our proposal does not require the widely used heredity assumption. In addition, our proposed estimates have explicit formulas and obey the invariance principle at the population level. We estimate the interactions of matrix form under penalized convex loss function. The resulting estimates are shown to be consistent even when the covariate dimension is an exponential order of the sample size. We develop an efficient ADMM algorithm to implement the penalized estimation. This ADMM algorithm fully explores the cheap computational cost of matrix multiplication and hence is much more efficient than existing penalized methods under heredity constraints. We demonstrate the promising performance of our proposal through extensive numerical studies.

报告人简介：蒋滨雁，2007年本科毕业于中国科学技术大学，2012年于新加坡国立大学获得博士学位，后在卡内基梅隆大学做过访问学者（博士后）。现任香港理工大学数学学院助理教授，研究成果发表在Biometrika、JASA、CSDA、Statistica Sinica等统计学顶级期刊。主要研究方向是复杂数据比如高维数据分析。

报告时间：2018年11月19日（星期一）上午8：30 -9：30.

报告地点：科技楼南楼702

(二)报告人：黄磊（西南交通大学）

报告题目：Weighted Volume Under the Three-way Receiver Operating Characteristic Surface

报告摘要：In medical researches, three-way ROC analysis has been widely concerned. This paper generalizes the volume under the surface (VUS) of three-way ROC analysis to weighted volume under the surface (WVUS) by introducing a weight function into the integration formula. This generalization is far from trivial; it practically allows researchers to calculate the diagnostic accuracy of some tests while some high probabilities of correct classification for certain classes are ensured. Theoretically, the asymptotic properties of the corresponding nonparametric and parametric estimators of WVUS have been well derived in this paper, which could provide some statistical inferences when comparing different diagnostic accuracies. Substantial simulations have been conducted to show the usefulness of the proposed estimators and also to show the necessity of considering WVUS. The applicability and feasibility of the proposed method has also been verified by a real data analysis about liver cancer.

报告人简介：黄磊，2010年本科毕业于中国科学技术大学，2015年博士毕业于新加坡国立大学，现任西南交通大学数学学院统计系助理教授，硕士生导师，主要研究方向有非参数、半参数时间序列模型，金融统计分析，医学生物统计。发表多篇SCI期刊文章，其中包括AOS, SMMR, CSDA和JSCS等。主持自然科学基金青年项目一项，参与面上项目、青年项目各一项，2017年度国家留学基金委访问学者，获西南交通大学“青苗计划”学者称号。

报告时间：2018年11月19日（星期一）上午9：30 -10：30.

报告地点：科技楼南楼702

（三）报告人：刘成（武汉大学）

报告题目：A Simple and Trustworthy Asymptotic t Test in Difference-in-Differences Regressions

报告摘要：The paper proposes an asymptotically valid t test in a difference-in-differences (DD) regression when the number of time periods is large while the number of individuals can be small or large. The proposed t test is based on a special heteroscedasticity and autocorrelation robust (HAR) variance estimator that is tailored to the inference problems in the DD setting. The asymptotic distribution of the t test depends on the smoothing parameter K in the HAR variance estimator, and a testing-optimal procedure for choosing K is developed through minimizing the type II error subject to a constraint on the type I error of the t test. By capturing the estimation uncertainty of the HAR variance estimator, the t test has more accurate size than the corresponding normal test and is just as powerful as the latter. Compared to the nonstandard test that is designed to reduce the size distortion of the normal test, the proposed t test is just as accurate but much more convenient to use, as the critical values are from the standard t table. Model-based and empirical-data-based Monte Carlo simulations show that the proposed t test works quite well in finite samples.

报告人简介：刘成，2009年毕业于武汉大学sunbet中国官网统计系，2013年于新加坡国立大学统计与应用概率系取得博士学位，现任职于武汉大学经济与管理学院副教授。已于计量经济学国际顶刊Journal of Econometrics发表文章2篇。

报告时间：2018年11月19日（星期一）上午10：30 -11：30.

报告地点：科技楼南楼702