Model selection and estimation in the Gaussian graphical model
The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index.
Abstract
We propose penalized likelihood methods for estimating the concentration matrix in the Gaussian graphical model. The methods lead to a sparse and shrinkage estimator of the concentration matrix that is positive definite, and thus conduct model selection and estimation simultaneously. The implementation of the methods is nontrivial because of the positive definite constraint on the concentration matrix, but we show that the computation can be done effectively by taking advantage of the efficient maxdet algorithm developed in convex optimization. We propose a BIC-type criterion for the selection of the tuning parameter in the penalized likelihood methods. The connection between our methods and existing methods is illustrated. Simulations and real examples demonstrate the competitive performance of the new methods.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“These methods have the advantage to be able to recover the off diagonal structure of the covariance matrix.”
From Signal Estimation and Uncertainties Extraction in Terahertz Time-Domain Spectroscopy (2025) · cites this paper as [58]
Citations
Cites
Boyd, S., & Vandenberghe, L. (2004) · doi:10.1017/cbo9780511804441 · Cited by 56,601
Regression shrinkage and selection via the lasso
Tibshirani, R. (1996) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1996.tb02080.x · Cited by 51,482 · cited 1×
“This is similar to the idea of the lasso in linear regression (Tibshirani, 1996).”
MethodGraphical Models in Applied Multivariate Statistics
Goodall, C. R. (1991) · Technometrics · doi:10.1080/00401706.1991.10484880 · Cited by 1,849
Asymptotics for lasso-type estimators
Fu, W., & Knight, K. (2000) · The Annals of Statistics · doi:10.1214/aos/1015957397 · Cited by 1,276 · cited 2×
“, X(p)) be a p-dimensional random vector following a multivariate normal distribution Np(µ, ) with unknown mean µ and nonsingular covariance matrix .”
BackgroundMethodInfluentialHeuristics of instability and stabilization in model selection
Breiman, L. (1996) · The Annals of Statistics · doi:10.1214/aos/1032181158 · Cited by 1,146 · cited 1×
“As demonstrated by Breiman (1996), the discrete nature of such procedures often leads to instability of the estimator: small changes in the data may result in very different estimates.”
BackgroundBetter Subset Regression Using the Nonnegative Garrote
Breiman, L. (1995) · Technometrics · doi:10.1080/00401706.1995.10484371 · Cited by 1,043 · cited 2×
“As demonstrated by Breiman (1996), the discrete nature of such procedures often leads to instability of the estimator: small changes in the data may result in very different estimates.”
BackgroundMethodIntroduction to Graphical Modelling
Edwards, D. (2000) · Springer Texts in Statistics · doi:10.1007/978-1-4612-0493-0 · Cited by 825
Determinant Maximization with Linear Matrix Inequality Constraints
Vandenberghe, L., Boyd, S., & Wu, S.-P. (1998) · SIAM Journal on Matrix Analysis and Applications · doi:10.1137/s0895479896303430 · Cited by 652 · cited 2×
“To use the algorithm of Vandenberghe et al. (1998), it is also required that Fi, i = 1,. .”
MethodMultivariate Dependencies: Models, Analysis and Interpretation.
Murphy, S. A., Cox, D. R., & Wermouth, N. (1998) · Biometrics · doi:10.2307/2534028 · Cited by 487
Sparsistency and rates of convergence in large covariance matrix estimation
Lam, C., & Fan, J. (2009) · The Annals of Statistics · doi:10.1214/09-aos720 · Cited by 418
Cited by
Sparse inverse covariance estimation with the graphical lasso
Friedman, J., Hastie, T., & Tibshirani, R. (2008) · Biostatistics · doi:10.1093/biostatistics/kxm045 · Cited by 6,355
“Other authors have proposed algorithms for the exact maximization of the L 1 -penalized log-likelihood; Yuan & Lin (2007), Banerjee et al. (2007) and Dahl et al. (2007) adapt interior point optimization methods for the solution to this problem.”
InfluentialRegression Shrinkage and Selection via The Lasso: A Retrospective
Tibshirani, R. (2011) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.1467-9868.2011.00771.x · Cited by 3,589
Meinshausen, N., & Bühlmann, P. (2010) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.1467-9868.2010.00740.x · Cited by 2,215
“For the lasso, it amounts to the ‘neighbourhood stability’ condition (Meinshausen and Bühlmann, 2006), which is equivalent to the ‘irrepresentable condition’ (Zhao and Yu, 2006; Zou, 2006; Yuan and Lin, 2007).”
BackgroundMethodInfluentialSparse and Compositionally Robust Inference of Microbial Ecological Networks
Kurtz, Z. D., Müller, C. L., Miraldi, E. R., Littman, D. R., Blaser, M. J., & Bonneau, R. A. (2015) · PLOS Computational Biology · doi:10.1371/journal.pcbi.1004226 · Cited by 1,799
“A wealth of theoretical results are available that characterize conditions for asymptotic and finite sample guarantees for the estimated networks [20, 29, 39, 43, 46].”
BackgroundMethodThe Gaussian Graphical Model in Cross-Sectional and Time-Series Data
Epskamp, S., Waldorp, L. J., Mõttus, R., & Borsboom, D. (2018) · Multivariate Behavioral Research · doi:10.1080/00273171.2018.1454823 · Cited by 1,134
“A particularly popular variant of LASSO is the graphical LASSO (GLASSO; Friedman et al., 2008), which directly penalizes elements of the inverse variance-covariance matrix (Witten, Friedman, & Simon, 2011; Yuan & Lin, 2007).”
MethodA Constrained ℓ1Minimization Approach to Sparse Precision Matrix Estimation
Cai, T., Liu, W., & Luo, X. (2011) · Journal of the American Statistical Association · doi:10.1198/jasa.2011.tm10155 · Cited by 973
“In particular, the 1 penalized normal likelihood estimator and its variants, which we call 1-MLE type estimators, have been considered by several authors (see, e.g., Yuan and Lin 2007; d’Aspremont, Banerjee, and El Ghaoui 2008; Friedman, Hastie, and Tibshirani 2008; Rothman et al. 2008).”
BackgroundMethodCovariance regularization by thresholding
Bickel, P. J., & Levina, E. (2008) · The Annals of Statistics · doi:10.1214/08-aos600 · Cited by 893
“The common approach of [5, 31, 35] is to add an L 1 (lasso) penalty on the entries of the concentration matrix to the normal likelihood, which results in shrinking some of the elements of the inverse to zero.”
BackgroundMethodHigh-Dimensional Statistics: A Non-Asymptotic Viewpoint
Wainwright, M. J. (2019) · Cambridge University Press eBooks · doi:10.1017/9781108627771 · Cited by 882
The Joint Graphical Lasso for Inverse Covariance Estimation Across Multiple Classes
Danaher, P., Wang, P., & Witten, D. M. (2014) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/rssb.12033 · Cited by 855
“A number of authors have instead taken a penalized log likelihood approach (Yuan & Lin 2007 b , Friedman, Hastie & Tibshirani 2007, Rothman et al. 2008): rather than maximizing (1.1), one can instead solve the problem maximize Θ { log det Θ − trace( SΘ ) − λ || Θ || 1 } , (1.2) where λ is a…”
BackgroundMethodInfluentialJones, D. T., Buchan, D. W. A., Cozzetto, D., & Pontil, M. (2012) · Bioinformatics · doi:10.1093/bioinformatics/btr638 · Cited by 770
High-dimensional covariance estimation by minimizing ℓ1-penalized log-determinant divergence
Ravikumar, P., Wainwright, M. J., Raskutti, G., & Yu, B. (2011) · Electronic Journal of Statistics · doi:10.1214/11-ejs631 · Cited by 744
“n > C1 d 2 (1 + 8 α )(2) ( τ log p+ log 4 ) (31)”
Partial Correlation Estimation by Joint Sparse Regression Models
Peng, J., Wang, P., Zhou, N., & Zhu, J. (2009) · Journal of the American Statistical Association · doi:10.1198/jasa.2009.0126 · Cited by 688
“In Yuan and Lin (2007), a BIC criterion is proposed for the penalized maximum likelihood approach.”
BackgroundMethodInfluentialLearning Laplacian Matrix in Smooth Graph Signal Representations
Dong, X., Thanou, D., Frossard, P., & Vandergheynst, P. (2016) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2016.2602809 · Cited by 640
“We believe that our new framework can benefit numerous emerging real world applications, such as the analysis of transportation, biomedical, and social networks, where it is critical to infer hidden relationships between data entities.”
BackgroundThe Nonparanormal: Semiparametric Estimation of High Dimensional Undirected Graphs
Liu, H., Lafferty, J., & Wasserman, L. (2018) · doi:10.1184/r1/6610712.v1 · Cited by 621
Sparse permutation invariant covariance estimation
Rothman, A. J., Bickel, P. J., Levina, E., & Zhu, J. (2008) · Electronic Journal of Statistics · doi:10.1214/08-ejs176 · Cited by 560
Estimation of (near) low-rank matrices with noise and high-dimensional scaling
Negahban, S., & Wainwright, M. J. (2011) · The Annals of Statistics · doi:10.1214/10-aos850 · Cited by 537
“focused on models with sparsity constraints, including the problem of sparse linear regression [45, 14, 17, 34, 9], banded or sparse covariance matrices [7, 8, 18], sparse inverse covariance matrices [49, 21, 42, 40], sparse eigenstructure [27, 2, 38], and sparse regression matrices [37, 30, 48, 25]. A theme common to…”
High-dimensional semiparametric Gaussian copula graphical models
Liu, H., Han, F., Yuan, M., Lafferty, J., & Wasserman, L. (2012) · The Annals of Statistics · doi:10.1214/12-aos1037 · Cited by 507
“Yuan and Lin (2007), Banerjee, Ghaoui and d’Aspremont (2008), and Friedman, Hastie and Tibshirani (2008) develop a penalized likelihood approach to directly estimate Ω. Rothman et al. (2008), Ravikumar et al. (2009a), and Lam and Fan (2009) study the theoretical properties of the penalized…”
BackgroundSelf-citationGeneralized Thresholding of Large Covariance Matrices
Rothman, A. J., Levina, E., & Zhu, J. (2009) · Journal of the American Statistical Association · doi:10.1198/jasa.2009.0101 · Cited by 489
“KEY WORDS: Covariance; High-dimensional data; Regularization; Thresholding; Sparsity”
BackgroundStability Approach to Regularization Selection (StARS) for High Dimensional Graphical Models
Liu, H., Roeder, K., & Wasserman, L. (2010) · PubMed · doi:10.48550/arxiv.1006.3316 · Cited by 554
“To handle this challenge, the graphical lasso or glasso [7, 24, 2] is rapidly becoming a popular method for estimating sparse undirected graphs.”
MethodInfluentialJoint estimation of multiple graphical models
Guo, J., Levina, E., Michailidis, G., & Zhu, J. (2011) · Biometrika · doi:10.1093/biomet/asq060 · Cited by 431
Sparsistency and rates of convergence in large covariance matrix estimation
Lam, C., & Fan, J. (2009) · The Annals of Statistics · doi:10.1214/09-aos720 · Cited by 418
Network exploration via the adaptive LASSO and SCAD penalties
Fan, J., Feng, Y., & Wu, Y. (2009) · The Annals of Applied Statistics · doi:10.1214/08-aoas215 · Cited by 397
“This penalized likelihood was later studied by d’Aspremont, Banerjee and Ghaoui (2008), Yuan and Lin (2007), Levina, Zhu and Rothman (2008), Rothman et al. (2008) and Friedman, Hastie and Tibshirani (2008) in the context of precision matrix estimation.”
MethodAn overview of the estimation of large covariance and precision matrices
Fan, J., Liao, Y., & Liu, H. (2016) · The Econometrics Journal · doi:10.1111/ectj.12061 · Cited by 385
Connecting the Dots: Identifying Network Structure via Graph Signal Processing
Mateos, G., Segarra, S., Marques, A. G., & Ribeiro, A. (2019) · IEEE Signal Processing Magazine · doi:10.1109/msp.2018.2890143 · Cited by 369
“As shown in [2], the resulting complexity for off-the-shelf interior point methods adopted in [64] is ( ) .”
InfluentialHigh Dimensional Inverse Covariance Matrix Estimation via Linear Programming
Yuan, M. (2010) · Journal of Machine Learning Research · doi:10.5555/1756006.1859930 · Cited by 468
Self-citationSparse Multivariate Regression With Covariance Estimation
Rothman, A. J., Levina, E., & Zhu, J. (2010) · Journal of Computational and Graphical Statistics · doi:10.1198/jcgs.2010.09188 · Cited by 350
“Imposing sparsity in B̂ for the purposes of identifying “master predictors” was proposed by Peng et al. (2010), who applied a combined constraint function C(B) = λC1(B) + (1 − λ)C2(B) for λ ∈ [0,1], where C1(B) = ∑j,k |bjk|, the lasso constraint (Tibshirani 1996) on the entries of B and C2(B) = ∑pj=1 (b2j1 + · · · + b…”
BackgroundInfluentialSparse estimation of a covariance matrix
Bien, J., & Tibshirani, R. J. (2011) · Biometrika · doi:10.1093/biomet/asr054 · Cited by 342
“More recently, Yuan & Lin (2007), Banerjee et al. (2008) and Friedman et al. (2007) frame this as a sparse estimation problem, performing penalized maximum likelihood with a lasso penalty on the inverse covariance matrix; this is known as the graphical lasso.”
MethodLearning brain connectivity of Alzheimer's disease by sparse inverse covariance estimation
Huang, S., Li, J., Sun, L., Ye, J., Fleisher, A., Wu, T., Chen, K., & Reiman, E. (2010) · NeuroImage · doi:10.1016/j.neuroimage.2009.12.120 · Cited by 340
“(1) and achieve the SICE in recent years (Yuan and Lin, 2007; Friedman et al., 2007; Levina et al., 2008; Li and Gui, 2006), including a method by us (Sun et al.”
BackgroundMethodInfluentialNew Insights and Faster Computations for the Graphical Lasso
Witten, D. M., Friedman, J. H., & Simon, N. (2011) · Journal of Computational and Graphical Statistics · doi:10.1198/jcgs.2011.11051a · Cited by 325
“Yuan and Lin (2007) proposed that rather than maximizing the log-likelihood (1.1), one should instead maximize the penalized log-likelihood over nonnegative definite matrices (cid”
BackgroundMethodInfluentialStrong oracle optimality of folded concave penalized estimation
Fan, J., Xue, L., & Zou, H. (2014) · The Annals of Statistics · doi:10.1214/13-aos1198 · Cited by 313
“The validation error of a generic estimator Θ̂ for Models 4 – 5 is defined as We computed the GLASSO and the CLIME by the R packages glasso and clime .”
BackgroundBootstrapping Lasso Estimators
Chatterjee, A., & Lahiri, S. N. (2011) · Journal of the American Statistical Association · doi:10.1198/jasa.2011.tm10159 · Cited by 293
Two-Sample Test of High Dimensional Means Under Dependence
Cai, T., Liu, W., & Xia, Y. (2013) · Journal of the Royal Statistical Society Series B (Statistical Methodology) · doi:10.1111/rssb.12034 · Cited by 280
“See, for example, Yuan and Lin (2007), Bickel and Levina (2008), Rothman et al. (2008), Ravikumar et al. (2008), Cai, Zhang and Zhou (2010), Yuan (2010), Cai and Liu (2011), Cai, Liu, and Luo (2011), and Cai and Yuan (2012).”
BackgroundThe graphical lasso: New insights and alternatives
Mazumder, R., & Hastie, T. (2012) · Electronic Journal of Statistics · doi:10.1214/12-ejs740 · Cited by 267
High-dimensional analysis of semidefinite relaxations for sparse principal components
Amini, A. A., & Wainwright, M. J. (2009) · The Annals of Statistics · doi:10.1214/08-aos664 · Cited by 308
“elaxation methods. Other work has tackled the problem of estimating sparse covariance matrices in the high-dimensional setting, using thresholding methods [3, 12] as well as ℓ1-regularization methods [8, 39]. A related problem—and the primary focus of this paper—is recovering sparse eigenvectors from high-dimensional…”
Likelihood-Based Selection and Sharp Parameter Estimation
Shen, X., Pan, W., & Zhu, Y. (2012) · Journal of the American Statistical Association · doi:10.1080/01621459.2011.645783 · Cited by 317
Regularized M-estimators with nonconvexity: Statistical and algorithmic theory for local optima
Loh, P., & Wainwright, M. J. (2013) · ScholarlyCommons (University of Pennsylvania) · Cited by 254
Sparse Inverse Covariance Matrix Estimation Using Quadratic Approximation
Hsieh, C., Sustik, M. A., Dhillon, I. S., & Ravikumar, P. (2013) · arXiv (Cornell University) · doi:10.48550/arxiv.1306.3212 · Cited by 251
“For further details on the background of !1 regularization in the context of GMRFs, we refer the reader to [20, 2, 8, 15].”
BackgroundFast alternating linearization methods for minimizing the sum of two convex functions
Goldfarb, D., Ma, S., & Scheinberg, K. (2013) · Mathematical Programming · doi:10.1007/s10107-012-0530-2 · Cited by 246
“…covariance matrix from sample data by solving the following maximum likelihood problem with an ℓ 1 -regularization term, (see e.g., [3, 18, 49, 52]). max { log det( X ) − Σ , X − ρ X 1 } , or equivalently, min {− log det( X ) + Σ , X + ρ X 1 } , (1.7) where ρ > 0 and Σ ∈ S n + (the set of…”
InfluentialRegularized estimation of large-scale gene association networks using graphical Gaussian models
Krämer, N., Schäfer, J., & Boulesteix, A.-L. (2009) · BMC Bioinformatics · doi:10.1186/1471-2105-10-384 · Cited by 237
“Since investigators are primarily interested in direct gene interactions, the GGM framework is attractive formodeling of regulatory networks: several recent methodological articles report successful applications of GGMs to the estimation of genetic networks from microarray data [5-10].”
Bayesian Graphical Lasso Models and Efficient Posterior Computation
Wang, H. (2012) · Bayesian Analysis · doi:10.1214/12-ba729 · Cited by 231
“The graphical lasso (Meinshausen and B¨uhlmann 2006; Yuan and Lin 2007; Banerjee et al. 2008; Friedman et al. 2008; Guo et al. 2011) is widely used for simultaneous graphical model determination and covariance matrix estimation.”
BackgroundMethodApproximation accuracy, gradient methods, and error bound for structured convex optimization
Tseng, P. (2010) · Mathematical Programming · doi:10.1007/s10107-010-0394-2 · Cited by 224
“with 0 ≤ λ < λ̄ ≤ ∞ and τ > 0, where 〈s, x〉 = trace[sx], I denotes the n × n identity matrix, and denotes the partial ordering with respect to the cone Sn + of positive semidefinite matrices [5,9,56,77,159].”
Sparse estimation of large covariance matrices via a nested Lasso penalty
Levina, E., Rothman, A., & Zhu, J. (2008) · The Annals of Applied Statistics · doi:10.1214/07-aoas139 · Cited by 214
“Here we use the Kullback–Leibler loss for the concentration matrix, which was used in Yuan and Lin (2007).”
BackgroundMethodDAGs with NO TEARS: Continuous Optimization for Structure Learning
Zheng, X., Aragam, B., Ravikumar, P., & Xing, E. P. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1803.01422 · Cited by 209
“Undirected graphical models, also known as Markov networks, is recognized as a convex problem (Yuan and Lin, 2007; Banerjee et al., 2008) nowadays, and hence can be solved using black-box convex optimizers such as CVX (Grant and Boyd, 2014).”
BackgroundStructure Learning in Graphical Modeling
Drton, M., & Maathuis, M. H. (2017) · Annual Review of Statistics and Its Application · doi:10.1146/annurev-statistics-060116-053803 · Cited by 209
“Yuan & Lin (2007) and Banerjee et al. (2008) proposed the graphical lasso ( glasso ) estimator where the minimization is over positive definite matrices K = ( κ vw ) ∈ R V × V and λ ≥ 0 is a tuning parameter.”
BackgroundAsymptotic normality and optimalities in estimation of large Gaussian graphical models
Ren, Z., Sun, T., Zhang, C.-H., & Zhou, H. H. (2015) · The Annals of Statistics · doi:10.1214/14-aos1286 · Cited by 206
“Yuan and Lin (2007) proposed to use the lasso penalty and studied its asymptotic properties when p is fixed. Ravikumar et al. (2011) derived the rate of convergence when the dimension p is high by applying a primal-dual witness”
BackgroundSparse High-Dimensional Models in Economics
Fan, J., Lv, J., & Qi, L. (2011) · Annual Review of Economics · doi:10.1146/annurev-economics-061109-080451 · Cited by 206
“…sparsity, factor model, or autoregressive model on the data to reduce the dimensionality (see, for example, Wu & Pourahmadi 2003, Huang et al. 2006, Yuan & Lin 2007, Bai & Ng 2008, Bickel & Levina 2008a,b, Fan et al. 2008, Levina et al. 2008, Rothman et al. 2008, Lam & Fan 2009, Cai et al. 2010).”
MethodIdentification of power distribution network topology via voltage correlation analysis
Bolognani, S., Bof, N., Michelotti, D., Muraro, R., & Schenato, L. (2013) · 52nd IEEE Conference on Decision and Control · doi:10.1109/cdc.2013.6760120 · Cited by 200
“Because of this a priori knowledge, we also don’t need the tools that have been developed for model selection [27] (i.”
MethodLeast angle and ℓ1 penalized regression: A review
Hesterberg, T., Choi, N. H., Meier, L., & Fraley, C. (2008) · Statistics Surveys · doi:10.1214/08-ss035 · Cited by 198
“More recently, Yuan and Lin (2007b) proved some consistency results and showed that the solution path is piecewise linear and hence can be computed efficiently.”
BackgroundMethodEstimating sparse precision matrix: Optimal rates of convergence and adaptive estimation
Cai, T. T., Liu, W., & Zhou, H. H. (2016) · The Annals of Statistics · doi:10.1214/13-aos1171 · Cited by 194
“Yuan and Lin (2007) proposed an ℓ1 penalized normal likelihood estimator and studied its theoretical properties.”
BackgroundBayesian Inference of Multiple Gaussian Graphical Models
Peterson, C., Stingo, F. C., & Vannucci, M. (2015) · Journal of the American Statistical Association · doi:10.1080/01621459.2014.896806 · Cited by 194
“In Gaussian graphical models (GGMs), also known as covariance selection models (Dempster 1972), the conditional independence relationships correspond to constraints on the precision matrix = −1 of the multivariate normal distribution xi ∼ N (μ, −1), i = 1, . . . , n, (2.1) with xi ∈ Rp the vector…”
InfluentialInformation-Theoretic Limits of Selecting Binary Graphical Models in High Dimensions
Santhanam, N. P., & Wainwright, M. J. (2012) · IEEE Transactions on Information Theory · doi:10.1109/tit.2012.2191659 · Cited by 193
CCLasso: correlation inference for compositional data through Lasso
Fang, H., Huang, C., Zhao, H., & Deng, M. (2015) · Bioinformatics · doi:10.1093/bioinformatics/btv349 · Cited by 191
“…is a very common assumption for under-determinated problems such as linear regression models (Tibshirani, 1996), Gaussian graphical models (Yuan and Lin, 2007) and compressed sensing (Candes and Tao, 2005) where the number of unknown parameters is larger, sometimes much larger, than the…”
BackgroundMethodCai, T. T., Ren, Z., & Zhou, H. H. (2016) · Electronic Journal of Statistics · doi:10.1214/15-ejs1081 · Cited by 190
“Yuan and Lin (2007) first proposed using Ω̂λ and studied its asymptotic properties for fixed p as n → ∞.”
BackgroundMethodInfluentialLoh, P., & Wainwright, M. J. (2013) · The Annals of Statistics · doi:10.1214/13-aos1162 · Cited by 205
“For multivariate Gaussian data, this program is an `1-regularized maximum likelihood estimate known as the graphical Lasso, and is a well-studied method for recovering the edge structure in a Gaussian graphical model [18, 19, 20].”
MethodRandom matrix theory in statistics: A review
Paul, D., & Aue, A. (2014) · Journal of Statistical Planning and Inference · doi:10.1016/j.jspi.2013.09.005 · Cited by 184
Review on statistical methods for gene network reconstruction using expression data
Wang, Y. X. R., & Huang, H. (2014) · Journal of Theoretical Biology · doi:10.1016/j.jtbi.2014.03.040 · Cited by 182
“A rich wealth of the literature exists on the problem of estimating sparse precision matrix in high dimensional GGMs (Zhou et al., 2011; Yuan and Lin, 2007; Friedman et al., 2007).”
BackgroundRegularized rank-based estimation of high-dimensional nonparanormal graphical models
Xue, L., & Zou, H. (2012) · The Annals of Statistics · doi:10.1214/12-aos1041 · Cited by 180
“Penalized likelihood methods have been studied for Gaussian graphical modeling [Yuan and Lin ( 2007)].”
MethodWang, M., Fang, E. X., & Liu, H. (2017) · Mathematical Programming · doi:10.1007/s10107-016-1017-3 · Cited by 178
“To induce sparsity, penalized regression methods have been well studied in the past decade (see [2,32,35] for examples), in which the component function h j ’s are assumed to be linearmost of the time.”
MethodCovariance Estimation: The GLM and Regularization Perspectives
Pourahmadi, M. (2011) · Statistical Science · doi:10.1214/11-sts358 · Cited by 176
“has been redicovered recently by Dégerine and Lambert-Lacroix (2003), Kurowicka and Cooke (2003) and Joe (2006), but its origin can be traced to a notable and somewhat neglected paper of Yule (1907), equation (25).”
BackgroundMethodInfluentialNode-Based Learning of Multiple Gaussian Graphical Models
Mohan, K., London, P., Fazel, M., Witten, D., & Lee, S. (2013) · PubMed · doi:10.48550/arxiv.1303.5145 · Cited by 176
“Consequently, a number of authors (among others, Yuan and Lin, 2007a; Friedman et al., 2007; Ravikumar et al., 2008; Banerjee et al., 2008; Scheinberg et al., 2010; Hsieh et al., 2011) have considered maximizing the penalized log likelihood maximize where S is the empirical covariance matrix, l is…”
BackgroundMethodInfluentialNetwork Inference via the Time-Varying Graphical Lasso
Hallac, D., Park, Y., Boyd, S., & Leskovec, J. (2017) · Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/3097983.3098037 · Cited by 174
“We first consider static inference, which is equivalent to the graphical lasso problem [2, 7, 35], and then build on it to extend to dynamic networks.”
BackgroundMethodInfluentialPenalized likelihood methods for estimation of sparse high-dimensional directed acyclic graphs
Shojaie, A., & Michailidis, G. (2010) · Biometrika · doi:10.1093/biomet/asq038 · Cited by 173
“Banerjee et al.(2008) and Friedman et al. (2008b) explored different aspects of the problem of estimating the concentration matrix using the lasso penalty, while Yuan and Lin(2007) andFan et al.(2007) considered other choices for the penalty.”
MethodFreijeiro‐González, L., Febrero–Bande, M., & González–Manteiga, W. (2021) · International Statistical Review · doi:10.1111/insr.12469 · Cited by 199
“Besides, for consistent variable selection using ˆ S L 1 = { j : ˆ β L 1 j = 0 } , it turns out that the design matrix of the model, X , needs to satisfy some assumptions.”
MethodA Survey on Sparse Learning Models for Feature Selection
Li, X., Wang, Y., & Ruiz, R. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2020.2982445 · Cited by 171
“Yuan and Lin [32] proved that the adaptive Lasso obtains consistent solutions and revealed that the solution path of adaptive Lasso is piecewise linear which can be computed efficiently.”
BackgroundMicrobiome Multi-Omics Network Analysis: Statistical Considerations, Limitations, and Opportunities
Jiang, D., Armour, C. R., Hu, C., Mei, M., Tian, C., Sharpton, T. J., & Jiang, Y. (2019) · Frontiers in Genetics · doi:10.3389/fgene.2019.00995 · Cited by 170
“There are two major statistical approaches for estimating the inverse covariance matrix in the high-dimensional Gaussian graphical model: the neighborhood selection method (Meinshausen and Bühlmann, 2006) and the graphical lasso method (Yuan and Lin, 2007; Friedman et al., 2008).”
MethodRegularized M-estimators with nonconvexity: Statistical and algorithmic theory for local optima
Loh, P., & Wainwright, M. J. (2013) · arXiv (Cornell University) · doi:10.48550/arxiv.1305.2436 · Cited by 169
“tries of Θ, and define Ω := Θ ∈ Rp×p| Θ = ΘT,Θ 0 . A host of statistical and algorithmic results have been established for the graphical Lasso in the case of Gaussian observations with an ℓ1-penalty [3, 8, 21, 25], and more recently for discrete-valued observations, as well [14]. In addition, a version of the graphical…”
BackgroundCVXR : An R Package for Disciplined Convex Optimization
Fu, A., Narasimhan, B., & Boyd, S. (2020) · Journal of Statistical Software · doi:10.18637/jss.v094.i14 · Cited by 203
“One way to estimate Σ is to maximize the log-likelihood with an l1-norm constraint (Yuan and Lin 2007; Banerjee, Ghaoui, and d’Aspremont 2008; Friedman, Hastie, and Tibshirani 2008), which amounts to the optimization problem maximize S log det(S)− tr(SQ) subject to S ∈ S+, ∑n i=1 ∑n j=1 |Sij | ≤ α.”
MethodEstimating time-varying networks
Kolar, M., Song, L., Ahmed, A., & Xing, E. P. (2010) · The Annals of Applied Statistics · doi:10.1214/09-aoas308 · Cited by 164
“…computationally intensive since finding the estimate involves solving a semidefinite program (SDP) and a number of authors have worked on finding a more efficient way to solve the problem (Banerjee et al., 2008; Yuan and Lin, 2007; Friedman et al., 2007b; Duchi et al., 2008a; Rothman et al., 2008).”
BackgroundConfidence intervals for high-dimensional inverse covariance estimation
Janková, J., & van de Geer, S. (2015) · Electronic Journal of Statistics · doi:10.1214/15-ejs1031 · Cited by 214
“A popular estimation procedure which relies on the sparsity in Θ ∗ is the graphical Lasso [5, 4, 16].”
MethodEstimation of Sparse Binary Pairwise Markov Networks using Pseudo-likelihoods.
Höfling, H., & Tibshirani, R. J. (2009) · Journal of Machine Learning Research · doi:10.5555/1577069.1577101 · Cited by 207
“Yuan and Lin (2007), Banerjee et al. (2008) and Dahl et al. (2008) as well as Friedman, Hastie, and Tibshirani (2008a) propose algorithms for solving this penalized log-likelihood with the procedure in Friedman, Hastie, and Tibshirani (2008a), the graphical lasso, being especially fast and…”
MethodGaussian graphical model estimation with false discovery rate control
Liu, W. (2013) · The Annals of Statistics · doi:10.1214/13-aos1169 · Cited by 160
“Hence (p(2) − p)/2 max{ ∑ 1≤i<j≤p I{|T̂ij| ≥ 2 √ log p}, 1} ≤ p 2 − p 2 1 max(cp, dp) (29)”
MethodHigh-dimensional covariance matrix estimation with missing observations
Lounici, K. (2014) · Bernoulli · doi:10.3150/12-bej487 · Cited by 159
“See [2, 7, 16, 27, 28, 38] and the references cited therein for more details.”
BackgroundSparse precision matrix estimation via lasso penalized D-trace loss
Zhang, T., & Zou, H. (2014) · Biometrika · doi:10.1093/biomet/ast059 · Cited by 180
“Yuan & Lin (2007) proposed the lasso penalized likelihood criterion and suggested using the maxdet algorithm to compute the estimator.”
BackgroundMethodEnhanced construction of gene regulatory networks using hub gene information
Yu, D., Lim, J., Wang, X., Liang, F., & Xiao, G. (2017) · BMC Bioinformatics · doi:10.1186/s12859-017-1576-1 · Cited by 149
Integration of ‘omics’ data in aging research: from biomarkers to systems biology
Zierer, J., Menni, C., Kastenmüller, G., & Spector, T. D. (2015) · Aging Cell · doi:10.1111/acel.12386 · Cited by 155
QUIC: quadratic approximation for sparse inverse covariance estimation
Hsieh, C., Sustik, M. A., Dhillon, I. S., & Ravikumar, P. (2014) · Journal of Machine Learning Research · Cited by 146
A new approach to Cholesky-based covariance regularization in high dimensions
Rothman, A. J., Levina, E., & Zhu, J. (2010) · Biometrika · doi:10.1093/biomet/asq022 · Cited by 141
“…matrix, a popular method is to add the lasso (ℓ1) penalty on the entries of the inverse covariance matrix to the normal likelihood (d’Aspremont et al., 2008; Yuan and Lin, 2007; Rothman et al., 2008; Friedman et al., 2008), which has been extended to more general penalties by Lam and Fan (2007).”
MethodEstimating functional brain networks by incorporating a modularity prior
Qiao, L., Zhang, H., Kim, M., Teng, S., Zhang, L., & Shen, D. G. (2016) · NeuroImage · doi:10.1016/j.neuroimage.2016.07.058 · Cited by 137
“One is L1-regularized maximum likelihood estimation [4, 21], also known as graphical LASSO [17]; another is L1-regularized linear regression, which shares the same model with sparse representation (SR) and traditional LASSO [22, 23].”
MethodExact covariance thresholding into connected components for large-scale Graphical LassoPreprint
Mazumder, R., & Hastie, T. J. (2011) · arXiv (Cornell University) · doi:10.48550/arxiv.1108.3829 · Cited by 132
“…1The first draft to this paper is available at http://arxiv.org/PS_cache/arxiv/pdf/1108/1108.3829v1.pdf. lasso [Friedman et al., 2007, Banerjee et al., 2008, Yuan and Lin, 2007] estimates the covariance matrix Σ, under the assumption that the inverse covariance matrix i.e. Σ−1 is sparse.”
BackgroundOptimal estimation and rank detection for sparse spiked covariance matrices
Cai, T., Ma, Z., & Wu, Y. (2014) · Probability Theory and Related Fields · doi:10.1007/s00440-014-0562-z · Cited by 130
“thods, including banding, tapering, thresholding and penalization, have been introduced for estimating several classes of covariance and precision matrices with different structures. See, for example, [3, 2, 6, 10, 7, 9, 8, 16, 20, 22, 26, 37, 43, 44, 45], among many others. In the present paper, we consider spiked cov…”
MethodVast volatility matrix estimation for high-frequency financial data
Wang, Y., & Zou, J. (2010) · The Annals of Statistics · doi:10.1214/09-aos730 · Cited by 129
Robust inference with knockoffs
Barber, R. F., Candès, E. J., & Samworth, R. J. (2020) · The Annals of Statistics · doi:10.1214/19-aos1852 · Cited by 168
Positive definite estimators of large covariance matrices
Rothman, A. J. (2012) · Biometrika · doi:10.1093/biomet/ass025 · Cited by 128
Large-scale Sparse Inverse Covariance Matrix Estimation
Bollhöfer, M., Eftekhari, A., Scheidegger, S., & Schenk, O. (2019) · SIAM Journal on Scientific Computing · doi:10.1137/17m1147615 · Cited by 127
“Among these there are blockwise descent methods [3,10,14,28], (inexect) interior point methods [4,21,35], alternating linearization [29], iterative thresholding [27], projected subgradients [11], greedy-type descent methods [30], and, more recently, second-order methods [2, 9, 17, 18, 25].”
BackgroundMethodDobra, A., Lenkoski, A., & Rodríguez, A. (2011) · Journal of the American Statistical Association · doi:10.1198/jasa.2011.tm10465 · Cited by 151
“…a popular tool for sparse estimation in both the statistics and machine learning literatures (Dobra et al. 2004; Meinshausen and Bühlmann 2006; Yuan and Lin 2007; Banerjee, El Ghaoui, and D’Aspremont 2008; Drton and Perlman 2008; Friedman, Hastie, and Tibshirani 2008; Ravikumar, Wainwright,…”
MethodDINGO: differential network analysis in genomics
Ha, M. J., Baladandayuthapani, V., & Do, K.-A. (2015) · Bioinformatics · doi:10.1093/bioinformatics/btv406 · Cited by 120
“We use the Bayesian information criterion (BIC) to select the tuning parameter (Yuan and Lin, 2007).”
MethodCovariate-adjusted precision matrix estimation with an application in genetical genomics
Cai, T., Li, H., Liu, W., & Xie, J. (2012) · Biometrika · doi:10.1093/biomet/ass058 · Cited by 145
“Estimation of the precision matrix and the graphical structure can also be obtained through a penalized maximum likelihood approach, see, for example, Friedman et al. (2008), Rothman et al. (2008) and Yuan & Lin (2007).”
Selection and estimation for mixed graphical models
Chen, S., Witten, D., & Shojaie, A. (2014) · Biometrika · doi:10.1093/biomet/asu051 · Cited by 139
“In particular, Gaussian graphical models have been studied extensively (see, e.g., Meinshausen & Bühlmann, 2006; Yuan & Lin, 2007; Friedman et al., 2008; Rothman et al., 2008; Wainwright & Jordan, 2008; Peng et al., 2009; Ravikumar et al., 2011) and generalized to account for nonnormality and…”
BackgroundMethodInfluentialSparse Inverse Covariance Selection via Alternating Linearization MethodsPreprint
Scheinberg, K., Ma, S., & Goldfarb, D. (2010) · arXiv (Cornell University) · doi:10.48550/arxiv.1011.0097 · Cited by 117
“Given a penalty parameter 1 /µ , at the k -th iteration, the augmented Lagrangian method minimizes the augmented Lagrangian function L ( x, y ; λ ) with respect to x and y , i.e., it solves the subproblem and updates the Lagrange multiplier λ via: Since minimizing L ( x, y ; λ ) with respect to x…”
MethodScaling It Up: Stochastic Search Structure Learning in Graphical Models
Wang, H. (2015) · Bayesian Analysis · doi:10.1214/14-ba916 · Cited by 116
“Classical methods for inducing sparsity often rely on penalized likelihood approaches (Banerjee et al., 2008; Yuan and Lin, 2007; Bien and Tibshirani, 2011; Wang, 2014).”
MethodGemini: Graph estimation with matrix variate normal instances
Zhou, S. (2014) · The Annals of Statistics · doi:10.1214/13-aos1187 · Cited by 142
“We use the R-package glasso [9] to compute the GLasso solution.”
MethodInfluentialAdaptive covariance matrix estimation through block thresholding
Cai, T., & Yuan, M. (2012) · The Annals of Statistics · doi:10.1214/12-aos999 · Cited by 113
“See, for example, Ledoit and Wolf (2004), Huang et al. (2006), Yuan and Lin (2007), Banerjee, El Ghaoui and This is an electronic reprint of the original article published by the Institute of Mathematical Statistics in The Annals of Statistics , 2012, Vol. 40, No. 4, 2014–2042.”
BackgroundSelf-citationCovariance Estimation in High Dimensions Via Kronecker Product Expansions
Tsiligkaridis, T., & Hero, A. O. (2013) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2013.2279355 · Cited by 113
“The main difficulty in maximum likelihood estimation of structured covariances is the nonconvex optimization problem that arises.”
BackgroundTIGER: A tuning-insensitive approach for optimally estimating Gaussian graphical models
Liu, H., & Wang, L.-E. (2017) · Electronic Journal of Statistics · doi:10.1214/16-ejs1195 · Cited by 139
“A related approach is to directly estimate Θ by penalizing the likelihood using the L1-penalty (Banerjee et al., 2008; Yuan and Lin, 2007; Friedman et al., 2008).”
MethodMachine Learning and Artificial Intelligence in the Multi-Omics Approach to Gut Microbiota
Rozera, T., Pasolli, E., Segata, N., & Ianiro, G. (2025) · Gastroenterology · doi:10.1053/j.gastro.2025.02.035 · Cited by 112
An inexact interior point method for L 1-regularized sparse covariance selection
Lu, L., & Toh, K.-C. (2010) · Mathematical Programming Computation · doi:10.1007/s12532-010-0020-6 · Cited by 124
“Next we consider a collection of problems considered in [11,44].”
BackgroundMethodQiao, X., Guo, S., & James, G. M. (2017) · Journal of the American Statistical Association · doi:10.1080/01621459.2017.1390466 · Cited by 139
“In Section 2 we propose a convex penalized criterion which has connections to both the graphical lasso (Yuan and Lin, 2007) and the group lasso (Yuan and Lin, 2006).”
BackgroundMethodInfluentialLarge covariance estimation through elliptical factor models
Fan, J., Liu, H., & Wang, W. (2018) · The Annals of Statistics · doi:10.1214/17-aos1588 · Cited by 145
“…possible methods can also be considered including graphical Lasso, graphical SCAD, graphical Dantzig selector, and graphical neighborhood selection (Friedman et al., 2008; Yuan and Lin, 2007; Fan et al., 2009; Lam and Fan, 2009; Ravikumar et al., 2011; Yuan, 2010; Meinshausen and B¨uhlmann, 2006).”
MethodLee, W., & Liu, Y. (2012) · Journal of Multivariate Analysis · doi:10.1016/j.jmva.2012.03.013 · Cited by 122
“For the asymptotic analysis, we use the set-up of Fan and Li [6], Yuan and Lin [23] and Zou [25].”
MethodInfluentialHigh Dimensional Semiparametric Latent Graphical Model for Mixed Data
Fan, J., Liu, H., Ning, Y., & Zou, H. (2016) · Journal of the Royal Statistical Society Series B (Statistical Methodology) · doi:10.1111/rssb.12168 · Cited by 146
“…to the mathematical simplicity and wide applicability, Gaussian graphical models have been extensively studied by Meinshausen and B¨uhlmann (2006); Yuan and Lin (2007); Rothman et al. (2008); Friedman et al. (2008); d’Aspremont et al. (2008); Fan et al. (2009); Lam and Fan (2009); Yuan (2010);…”
Background
Metadata
- Authors
- Yuan, M., & Lin, Y.
- Affiliations
- Georgia Institute of Technology (US)
- University of Wisconsin–Madison (US)
- Venue
- Biometrika
- Publisher
- Oxford University Press (OUP)
- Published
- 2007-02-07
- Volume
- 94
- Issue
- 1
- Pages
- 19-35
- ISSN
- 0006-3444 (ISSN-L)
- 1464-3510
- OpenAlex
- W2081746825
- Semantic Scholar
- 20ee888ca9c3bae784f266405b4d6395c6f32955
- Cited by
- 1,627
- References
- 16
- Citations by year
- 2026: 43 · 2025: 66 · 2024: 60 · 2023: 64 · 2022: 86 · 2021: 131 · 2020: 112 · 2019: 147 · 2018: 142 · 2017: 91 · 2016: 123 · 2015: 98 · 2014: 109 · 2013: 86 · 2012: 85
- FWCI
- 38.99
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Biometrika (publishedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- HighWire Press Open Archive (submittedVersion)
- Funders
- National Science Foundation
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Inference
- Physical Sciences › Mathematics › Statistics and Probability › Random Matrices and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Modeling and Causal Inference
- Keywords
- Gaussian graphical models, model selection, penalized likelihood, sparse estimation, shrinkage estimation, hyperparameter optimization, covariance structure
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.