Estimating sparse networks with hubs

DOI10.1016/J.JMVA.2020.104655zbMATH Open1454.62181arXiv1904.09394OpenAlexW3041970402MaRDI QIDQ2196140FDOQ2196140

Authors: Annaliza McGillivray, Abbas Khalili, David A. Stephens 

Publication date: 28 August 2020

Published in: Journal of Multivariate Analysis (Search for Journal in Brave)

Abstract: Graphical modelling techniques based on sparse selection have been applied to infer complex networks in many fields, including biology and medicine, engineering, finance, and social sciences. One structural feature of some of the networks in such applications that poses a challenge for statistical inference is the presence of a small number of strongly interconnected nodes in a network which are called hubs. For example, in microbiome research hubs or microbial taxa play a significant role in maintaining stability of the microbial community structure. In this paper, we investigate the problem of estimating sparse networks in which there are a few highly connected hub nodes. Methods based on L1-regularization have been widely used for performing sparse selection in the graphical modelling context. However, while these methods encourage sparsity, they do not take into account structural information of the network. We introduce a new method for estimating networks with hubs that exploits the ability of (inverse) covariance selection methods to include structural information about the underlying network. Our proposed method is a weighted lasso approach with novel row/column sum weights, which we refer to as the hubs weighted graphical lasso. We establish large sample properties of the method when the number of parameters diverges with the sample size, and evaluate its finite sample performance via extensive simulations. We illustrate the method with an application to microbiome data.

Full work available at URL: https://arxiv.org/abs/1904.09394

Recommendations

• Sparse inverse covariance estimation with the graphical lasso
• Network exploration via the adaptive LASSO and SCAD penalties
• Learning graphical models with hubs
• Sparse estimation of conditional graphical models with application to gene networks
• Sparse estimation of a covariance matrix

zbMATH Keywords

Gaussian graphical modelsparsityhubsweighted Lasso

Mathematics Subject Classification ID

Asymptotic properties of parametric estimators (62F12) Applications of statistics to biology and medical sciences; meta analysis (62P10) Gaussian processes (60G15) Estimation in multivariate analysis (62H12) Probabilistic graphical models (62H22) Ridge regression; shrinkage estimators (Lasso) (62J07)

Cites Work

• The Adaptive Lasso and Its Oracle Properties
• Variable Selection via Nonconcave Penalized Likelihood and its Oracle Properties
• High-dimensional graphs and variable selection with the Lasso
• A unified framework for high-dimensional analysis of \(M\)-estimators with decomposable regularizers
• Sparsistency and rates of convergence in large covariance matrix estimation
• Sparse inverse covariance estimation with the graphical lasso
• Sparse permutation invariant covariance estimation
• Network exploration via the adaptive LASSO and SCAD penalties
• Regularized estimation of large covariance matrices
• Emergence of Scaling in Random Networks
• Model selection and estimation in the Gaussian graphical model
• Likelihood-based selection and sharp parameter estimation
• Tuning parameter selection for penalized likelihood estimation of Gaussian graphical model
• Consistency of the group Lasso and multiple kernel learning
• An overview of the estimation of large covariance and precision matrices
• Node-based learning of multiple Gaussian graphical models
• Hub Discovery in Partial Correlation Graphs
• Large covariance estimation for compositional data via composition-adjusted thresholding
• Learning graphical models with hubs

Cited In (8)

• Estimation of Sparse Directional Connectivity With Expectation Maximization
• Estimating scale-free networks via the exponentiation of minimax concave penalty
• Detection of hubs in complex networks by the Laplacian matrix
• Maximum likelihood estimation of sparse networks with missing observations
• Low-rank matrix estimation via nonconvex optimization methods in multi-response errors-in-variables regression
• Analysis of Networks via the Sparseβ-model
• Learning graphical models with hubs
• Robust methods for inferring sparse network structures

Uses Software

• R
• igraph
• glasso
• hglasso

This page was built for publication: Estimating sparse networks with hubs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2196140)