题 目：Some Results on Penalized Regression for Distributed Data

报告人：李楚进 博 士

时 间：2020年11月28日（星期六）上午9:00-11:00

地点：腾讯会议：324 438 768

内 容：In modern statistic problems, researchers are confronted with the challenges of the high-dimensional data, which are randomly divided across many machines. We develop the novel and communication-efficient approach for sparse and high-dimensional data with the penalized expectile or quantile regression. In each round, our method only requires the master machine to deal with a sparse penalized expectile or quantile regression which can be computed fastly by proximal alternating direction method of multipliers (ADMM) algorithm, and the other worker machines to compute the sub-gradient on local data. The advantage of the proximal ADMM algorithm is that it can make every parameter of iteration have closed formula in high-dimensional case, which greatly improves the speed of calculation. As for the communication efficiency, the proposed approach does not sacrifice any statistical accuracy and provably improves the estimation error obtained by centralized method, provided the penalty levels are chosen properly. And, we present extensive experiments on both the numerical simulations and practical data analysis, which all confirm the significant efficiency of our proposed method in expectile or quantile regression for distributed data.

报告人简介：李楚进，博士，博士后，华中科技大学数学与统计学院副教授，主要研究方向为随机分析、数理统计与应用统计。曾主持教育部留学回国人员启动基金项目、湖北省统计科研计划项目，参与多项国家自然科学基金项目以及省级和校级教研教改项目等；曾获得湖北省教学成果一等奖，发表了多篇SCI科研论文与教研教改论文。