Inference post Selection of Group-sparse Regression Models

12/31/2020
by   Snigdha Panigrahi, et al.
11

Conditional inference provides a rigorous approach to counter bias when data from automated model selections is reused for inference. We develop in this paper a statistically consistent Bayesian framework to assess uncertainties within linear models that are informed by grouped sparsities in covariates. Finding wide applications when genes, proteins, genetic variants, neuroimaging measurements are grouped respectively by their biological pathways, molecular functions, regulatory regions, cognitive roles, these models are selected through a useful class of group-sparse learning algorithms. An adjustment factor to account precisely for the selection of promising groups, deployed with a generalized version of Laplace-type approximations is the centerpiece of our new methods. Accommodating well known group-sparse models such as those selected by the Group LASSO, the overlapping Group LASSO, the sparse Group LASSO etc., we illustrate the efficacy of our methodology in extensive experiments and on data from a human neuroimaging application.

READ FULL TEXT

page 9

page 30

page 31

page 32

page 33

research
02/18/2014

Classification with Sparse Overlapping Groups

Classification with a sparsity constraint on the solution plays a centra...
research
06/24/2023

Selective inference using randomized group lasso estimators for general models

Selective inference methods are developed for group lasso estimators for...
research
03/29/2022

A new procedure for Selective Inference with the Generalized Linear Lasso

This articles investigates the distribution of the solutions of the gene...
research
12/21/2021

Group Lasso merger for sparse prediction with high-dimensional categorical data

Sparse prediction with categorical data is challenging even for a modera...
research
09/12/2017

Identifying Genetic Risk Factors via Sparse Group Lasso with Group Graph Structure

Genome-wide association studies (GWA studies or GWAS) investigate the re...
research
10/24/2019

Joint User Selection and Precoding in Multiuser MIMO Systems via Group LASSO

Joint user selection and precoding in multiuser MIMO settings can be int...
research
06/03/2022

Multivariate Sparse Group Lasso Joint Model for Radiogenomics Data

Radiogenomics is an emerging field in cancer research that combines medi...

Please sign up or login with your details

Forgot password? Click here to reset