高维线性模型中的纠偏LASSO综述

Overview of Debiased LASSO in High-Dimensional Linear Model

我们对高维线性模型统计推断近年来发展的一些结果进行了综述.我们介绍了三种主要的纠偏LASSO估计的思想与方法,这些纠偏LASSO估计具有渐近正态性,因此可以对低维系数进行统计推断.此外,我们也对基于纠偏LASSO进行bootstrap抽样的同时推断方法做了简单介绍.

We review some results on the recent development of statistical inference for high-dimensional linear models.We introduce three debiased LASSO estimators,which are asymptotically normal and thus we can construct statistical inference for low dimensional parameters in high-dimensional setting.In addition,we give a brief introduction to the bootstrap assistant procedures to conduct simultaneous inference based on the debiased LASSO.