021-55664888

扫描二维码关注
复旦大学管理学院

查看更多

统计学系系列讲座之312期

 

时间:2018年7月4日(周三)15:00-16:00

地点:史带楼801室

主持人:肖志国 副教授 复旦大学管理学院统计学系

主题:High-dimensional Cost-constrained Regression via Non-convex Optimization

主讲人:Prof. Yufeng Liu   University of North Carolina at Chapel Hill

 

简介:刘玉峰教授,目前担任Journal of Royal Statistical Society, Series B.和 Journal of Multivariate Analysis 的Associate Editor。他2013年当选ASA的Fellow,2017年当选IMS的Fellow。他的研究领域为:Statistical Machine Learning and Data Mining,High-dimensional Data Analysis,Nonparametric Statistics and Functional Estimation,Bioinformatics等。

 

摘要:In modern predictive modeling process, budget constraints become a very important consideration due to the high cost of collecting data using new techniques such as brain imaging and DNA sequencing. This motivates us to develop new and efficient high-dimensional cost constrained predictive modeling methods. In this talk, to address this challenge, we first study a new non-convex high-dimensional cost-constrained linear regression problem, that is, to find the cost-constrained regression model with the smallest expected prediction error among all models satisfying a budget constraint. The non-convex budget constraint makes this problem NP-hard. In order to estimate the regression coefficient vector of the cost-constrained regression model, we propose a new discrete extension of recent first-order continuous optimization methods. Theoretically, we prove that the series of the estimates generated by our iterative algorithm converge to a first-order stationary point, which can be a globally optimal solution to the nonconvex high-dimensional cost-constrained regression problem. Computationally, our numerical studies show that the proposed method can solve problems of fairly high dimensions and has promising estimation, prediction, and model selection performance.

 

 

                                                           统计学系 

2018-6-29

 

快速链接

扫一扫,欢迎关注
复旦管院官方微信