The graphical lasso: new insights and alternatives

DOI10.1214/12-EJS740zbMATH Open1295.62066arXiv1111.5479OpenAlexW2105760337WikidataQ41957694 ScholiaQ41957694MaRDI QIDQ1950894FDOQ1950894

Authors: Rahul Mazumder, Trevor Hastie 

Publication date: 28 May 2013

Published in: Electronic Journal of Statistics (Search for Journal in Brave)

Abstract: The graphical lasso citep{FHT2007a} is an algorithm for learning the structure in an undirected Gaussian graphical model, using $el{l}_1$ regularization to control the number of zeros in the precision matrix $BTheta={BSigma}^{-1}$ citep{BGA2008,yuan_lin_07}. The { exttt R} package GL citep{FHT2007a} is popular, fast, and allows one to efficiently build a path of models for different values of the tuning parameter. Convergence of GL can be tricky; the converged precision matrix might not be the inverse of the estimated covariance, and occasionally it fails to converge with warm starts. In this paper we explain this behavior, and propose new algorithms that appear to outperform GL. By studying the "normal equations" we see that, GL is solving the {em dual} of the graphical lasso penalized likelihood, by block coordinate ascent; a result which can also be found in cite{BGA2008}. In this dual, the target of estimation is $BSigma$, the covariance matrix, rather than the precision matrix $BTheta$. We propose similar primal algorithms PGL and DPGL, that also operate by block-coordinate descent, where $BTheta$ is the optimization target. We study all of these algorithms, and in particular different approaches to solving their coordinate sub-problems. We conclude that DPGL is superior from several points of view.

Full work available at URL: https://arxiv.org/abs/1111.5479

Recommendations

• Sparse inverse covariance estimation with the graphical lasso
• Graphical Lasso and thresholding: equivalence and closed-form solutions
• Coordinate descent algorithm for covariance graphical Lasso
• A note on the lack of symmetry in the graphical lasso
• Bayesian graphical Lasso models and efficient posterior computation

zbMATH Keywords

sparsityprecision matrixsemidefinite programmingpositive definite matricesgraphical lassoconvex analysis/optimizationsparse inverse covariance selection

Mathematics Subject Classification ID

Multivariate analysis (62H99) Learning and adaptive systems in artificial intelligence (68T05) Software, source code, etc. for problems pertaining to statistics (62-04) Graphical methods in statistics (62A09) Semidefinite programming (90C22)

Cites Work

• A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
• Title not available (Why is that?)
• High-dimensional graphs and variable selection with the Lasso
• Sparse inverse covariance estimation with the graphical lasso
• Sparse permutation invariant covariance estimation
• Model selection through sparse maximum likelihood estimation for multivariate Gaussian or binary data
• Model selection and estimation in the Gaussian graphical model
• Exact covariance thresholding into connected components for large-scale graphical lasso

Cited In (50)

• A focused information criterion for graphical models in fMRI connectivity with high-dimensional data
• An adapted linear discriminant analysis with variable selection for the classification in high-dimension, and an application to medical data
• Functional Graphical Models
• Title not available (Why is that?)
• Sparse seasonal and periodic vector autoregressive modeling
• Adjusted regularization in latent graphical models: application to multiple-neuron spike count data
• High-dimensional covariance matrix estimation
• Adjusted regularization of cortical covariance
• Incorporating grouping information into Bayesian Gaussian graphical model selection
• An efficient algorithm for sparse inverse covariance matrix estimation based on dual formulation
• A component Lasso
• Bayesian regularization for graphical models with unequal shrinkage
• Support recovery of Gaussian graphical model with false discovery rate control
• High-dimensional tests for functional networks of brain anatomic regions
• An efficient parallel block coordinate descent algorithm for large-scale precision matrix estimation using graphics processing units
• pISTA: preconditioned Iterative Soft Thresholding Algorithm for Graphical Lasso
• Graph informed sliced inverse regression
• Sparse estimation of high-dimensional inverse covariance matrices with explicit eigenvalue constraints
• Title not available (Why is that?)
• Certifiably optimal sparse inverse covariance estimation
• Estimation of graphical models: an overview of selected topics
• Equivalence of graphical Lasso and thresholding for sparse graphs
• Change-Point Detection for Graphical Models in the Presence of Missing Values
• The conditional censored graphical Lasso estimator
• A 2-stage elastic net algorithm for estimation of sparse networks with heavy-tailed data
• Gaussian graphical models with applications to omics analyses
• A Laplacian approach to \(\ell_1\)-norm minimization
• An Expectation Conditional Maximization Approach for Gaussian Graphical Models
• glasso
• Gaussian-related undirected graphical models for circular variables
• Fused multiple graphical lasso
• IDGM: an approach to estimate the graphical model of interval-valued data
• Honest confidence regions and optimality in high-dimensional precision matrix estimation
• Bayesian inference in nonparanormal graphical models
• Confidence graphs for graphical model selection
• Stretchy binary classification
• Fixed support positive-definite modification of covariance matrix estimators via linear shrinkage
• A proximal point dual Newton algorithm for solving group graphical Lasso problems
• Graphical lassos for meta‐elliptical distributions
• An efficient GPU-parallel coordinate descent algorithm for sparse precision matrix estimation via scaled Lasso
• Regularized multivariate regression models with skew-\(t\) error distributions
• Exact test theory in Gaussian graphical models
• Title not available (Why is that?)
• Graphical Lasso and thresholding: equivalence and closed-form solutions
• A note on the lack of symmetry in the graphical lasso
• Sparse reduced-rank regression with covariance estimation
• Bayesian analysis of nonparanormal graphical models using rank-likelihood
• Comparing dependent undirected Gaussian networks
• A unified framework for structured graph learning via spectral constraints
• Post-processed posteriors for sparse covariances

Uses Software

• R
• ElemStatLearn

This page was built for publication: The graphical lasso: new insights and alternatives

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1950894)