Abstract:In this paper, we propose two new algorithms for maximum-likelihood estimation (MLE) of high dimensional sparse covariance matrices. Unlike most of the state of-the-art methods, which either use regularization techniques or penalize the likelihood to impose sparsity, we solve the MLE problem based on an estimated covariance graph. More specifically, we propose a two-stage procedure: in the first stage, we determine the sparsity pattern of the target covariance matrix (in other words the marginal independence in the covariance graph under a Gaussian graphical model) using the multiple hypothesis testing method of false discovery rate (FDR), and in the second stage we use either a block coordinate descent approach to estimate the non-zero values or a proximal distance approach that penalizes the distance between the estimated covariance graph and the target covariance matrix. Doing so gives rise to two different methods, each with its own advantage: the coordinate descent approach does not require tuning of any hyper-parameters, whereas the proximal distance approach is computationally fast but requires a careful tuning of the penalty parameter. Both methods are effective even in cases where the number of observed samples is less than the dimension of the data. For performance evaluation, we test the proposed methods on both simulated and real-world data and show that they provide more accurate estimates of the sparse covariance matrix than two state-of-the-art methods.

Maximum Likelihood Estimation for Nets of Conics

Maximum likelihood degree of the two-dimensional linear Gaussian covariance model

Rational Maximum Likelihood Estimators of Kronecker Covariance Matrices

Likelihood Geometry of Reflexive Polytopes

Principal binets

Estimating linear covariance models with numerical nonlinear algebra

Maximum Likelihood for Matrices with Rank Constraints

Maximum Likelihood Estimation of Toric Fano Varieties

Two new algorithms for maximum likelihood estimation of sparse covariance matrices with applications to graphical modeling

Conic Sparsity: Estimation of Regression Parameters in Closed Convex Polyhedral Cones

Homaloidal Polynomials and Gaussian Models of Maximum Likelihood Degree One

Varieties with maximum likelihood degree one

Likelihood Geometry of the Squared Grassmannian

Confocal conics and 4-webs of maximal rank

Gaussian graphical models with toric vanishing ideals

On the maximum likelihood degree for Gaussian graphical models

Pseudo-likelihood-based $M$-estimation of random graphs with dependent edges and parameter vectors of increasing dimension

Concurrent normals problem for convex polytopes and Euclidean distance degree

Maximum likelihood thresholds of Gaussian graphical models and graphical lasso

Convex hulls of monomial curves, and a sparse positivstellensatz