报告题目：Multi Screen Penalty in High Dimensional Data
报 告 人：杨玥含
中央财经大学
地 点：统计研究院426教室
摘 要：We propose a multi-step method in high-dimensional sparse linear regression model. This method selects the variables by iterations to ameliorate the bias problem induced by the L1 penalty. High precision estimator is obtained by this method. We study the situation where the general one step method needs more strict assumptions to select the true model, and compare the estimation and model selection precision with them. Both theoretical results and simulations demonstrate the effectiveness of our procedure.
欢迎广大师生参加！
统计研究院
2017年11月27日