Large-Scale Sparse Inverse Covariance Estimation via Thresholding and Max-Det Matrix Completion

被引:0
|
作者
Zhang, Richard Y. [1 ]
Fattahi, Salar [1 ]
Sojoudi, Somayeh [2 ]
机构
[1] Univ Calif Berkeley, Dept Ind Engn & Operat Res, Berkeley, CA 94720 USA
[2] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Berkeley, CA 94720 USA
关键词
SELECTION; GRAPHS; MODEL;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The sparse inverse covariance estimation problem is commonly solved using an l(1)-regularized Gaussian maximum likelihood estimator known as "graphical lasso", but its computational cost becomes prohibitive for large data sets. A recent line of results showed-under mild assumptions-that the graphical lasso estimator can be retrieved by soft-thresholding the sample covariance matrix and solving a maximum determinant matrix completion (MDMC) problem. This paper proves an extension of this result, and describes a Newton-CG algorithm to efficiently solve the MDMC problem. Assuming that the thresholded sample covariance matrix is sparse with a sparse Cholesky factorization, we prove that the algorithm converges to an epsilon-accurate solution in O(n log(1/epsilon)) time and O(n) memory. The algorithm is highly efficient in practice: we solve the associated MDMC problems with as many as 200,000 variables to 7-9 digits of accuracy in less than an hour on a standard laptop computer running MATLAB.
引用
收藏
页数:10
相关论文
共 50 条
  • [21] Large Covariance Estimation for Compositional Data Via Composition-Adjusted Thresholding
    Cao, Yuanpei
    Lin, Wei
    Li, Hongzhe
    JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2019, 114 (526) : 759 - 772
  • [22] Estimation of sparse covariance matrix via non-convex regularization
    Wang, Xin
    Kong, Lingchen
    Wang, Liqun
    JOURNAL OF MULTIVARIATE ANALYSIS, 2024, 202
  • [23] CARPool covariance: fast, unbiased covariance estimation for large-scale structure observables
    Chartier, Nicolas
    Wandelt, Benjamin D.
    MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY, 2022, 509 (02) : 2220 - 2233
  • [24] Topology identification of undirected consensus networks via sparse inverse covariance estimation
    Hassan-Moghaddam, Sepideh
    Dhingra, Neil K.
    Jovanovic, Mihailo R.
    2016 IEEE 55TH CONFERENCE ON DECISION AND CONTROL (CDC), 2016, : 4624 - 4629
  • [25] High Dimensional Inverse Covariance Matrix Estimation via Linear Programming
    Yuan, Ming
    JOURNAL OF MACHINE LEARNING RESEARCH, 2010, 11 : 2261 - 2286
  • [26] Parallel stochastic gradient algorithms for large-scale matrix completion
    Recht B.
    Ré C.
    Mathematical Programming Computation, 2013, 5 (2) : 201 - 226
  • [27] SPARSE ESTIMATION OF LARGE COVARIANCE MATRICES VIA A NESTED LASSO PENALTY
    Levina, Elizaveta
    Rothman, Adam
    Zhu, Ji
    ANNALS OF APPLIED STATISTICS, 2008, 2 (01): : 245 - 263
  • [28] Improved Bounded Matrix Completion for Large-Scale Recommender Systems
    Fang, Huang
    Zhen, Zhang
    Shao, Yiqun
    Hsieh, Cho-Jui
    PROCEEDINGS OF THE TWENTY-SIXTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2017, : 1654 - 1660
  • [29] BYZANTINE-RESILIENT DISTRIBUTED LARGE-SCALE MATRIX COMPLETION
    Lin, Feng
    Ling, Qing
    Xiong, Zhiwei
    2019 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2019, : 8167 - 8171
  • [30] Evaluation of thresholding methods for activation likelihood estimation meta-analysis via large-scale simulations
    Frahm, Lennart
    Cieslik, Edna C.
    Hoffstaedter, Felix
    Satterthwaite, Theodore D.
    Fox, Peter T.
    Langner, Robert
    Eickhoff, Simon B.
    HUMAN BRAIN MAPPING, 2022, 43 (13) : 3987 - 3997