2016/10/31 by Qing Zhou, Seunghyun Min
Mathematics · #Advanced Statistical Methods and Models #Bias of an estimator #Block (permutation group theory) #Estimator #Inference #Invariant estimator #Minimum-variance unbiased estimator #Monte Carlo method #Norm (philosophy) #Parametric statistics #Random Matrices and Applications #Statistical Methods and Inference #Statistical inference #math.ST #stat.ME #stat.TH
paper · pdf · doi:10.1214/17-ejs1309
published as Electronic Journal of Statistics, Vol. 11 (2017): 3039-3080 · 35 pages, 4 figures
openalex created_date 2016/11/04 · arxiv created 2016/12/23 · openalex publication_date 2017/01/01 · arxiv updated 2017/08/16 · openalex updated_date 2026/08/05
To make statistical inference about a group of parameters on high-dimensional data, we develop the method of estimator augmentation for the block lasso, which is defined via block norm regularization. By augmenting a block lasso estimator β with the subgradient S of the block norm evaluated at β , we derive a closed-form density for the joint distribution of (β ,S) under a high-dimensional setting. This allows us to draw from an estimated sampling distribution of β , or more generally any function of (β ,S), by Monte Carlo algorithms. We demonstrate the application of estimator augmentation in group inference with the group lasso and a de-biased group lasso constructed as a function of (β ,S). Our numerical results show that importance sampling via estimator augmentation can be orders of magnitude more efficient than parametric bootstrap in estimating tail probabilities for significance tests. This work also brings new insights into the geometry of the sample space and the solution uniqueness of the block lasso. To broaden its application, we generalize our method to a scaled block lasso, which estimates the error variance simultaneously.