A Study on Bias Effect on Model Selection Criteria in Graphical Lasso

다국어 초록 (Multilingual Abstract)

Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an 1 A study on Bias Effect on Model Selection Criteria in Graphical Lasso Young-Geun Choi1, Seyoung Jeong2, Donghyeon Yu2* 1SK Telecom, Seoul 04539, Korea 2Department of Statistics, Inha University, Incheon 22212, Korea (Received Oct 8, 2018; Revised Nov 4, 2018; Accepted Nov 12, 2018) Abstract Graphical lasso is one of the most popular methods to estimate a sparse precision matrix, which is an inverse of a covariance matrix. The objective function of graphical lasso imposes an ℓ-penalty on the (vectorized) precision matrix, where a tuning parameter controls the strength of the penalization. The selection of the tuning parameter is practically and theoretically important since the performance of the estimation depends on an appropriate choice of tuning parameter. While information criteria (e.g. AIC, BIC, or extended BIC) have been widely used, they require an asymptotically unbiased estimator to select optimal tuning parameter. Thus, the biasedness of the ℓ- regularized estimate in the graphical lasso may lead to a suboptimal tuning. In this paper, we propose a two-staged bias-correction procedure for the graphical lasso, where the first stage runs the usual graphical lasso and the second stage reruns the procedure with an additional constraint that zero estimates at the first stage remain zero

참고문헌 (Reference)

1 Fan Y, "Tuning parameter selection in high dimensional penalized likelihood" 75 : 531-552, 2013

2 Zhao YJ, "Tumor markers for hepatocellular carcinoma" 1 : 593-598, 2013

3 Danaher P, "The joint graphical lasso for inverse covariance estimation across multiple classes" 76 : 373-397, 2014

4 Mazumder R, "The graphical lasso: new insights and alternatives" 6 : 2125-2149, 2012

5 Yu D, "Statistical completion of a partially identified graph with applications for the estimation of gene regulatory networks" 16 : 670-685, 2015

6 Friedman J, "Sparse inverse covariance estimation with the graphical lasso" 9 : 432-441, 2008

7 김재희, "Review of Connectivity and Dynamics of Neural Information Processing" 자연과학연구소 36 (36): 97-103, 2017

8 Tomida S, "Relapse-related molecular signature in lung adenocarcinomas identifies patients with dismal prognosis" 27 : 2793-2799, 2009

9 Peng J, "Partial correlation estimation by joint sparse regression models" 104 : 735-746, 2009

10 Witten D, "New insights and faster computations for the graphical lasso" 20 : 892-900, 2011

11 Vihinen P, "Matrix metalloproteinases in cancer:prognostic markers and therapeutic targets" 99 : 157-166, 2002

12 Belloni A, "Least squares after model selection in high-dimensional sparse models" 19 : 521-547, 2013

13 Oh JH, "Inference of radio-responsive gene regulatory networks using the graphical lasso algorithm" 15 : S5-, 2014

14 Yu H, "Identification and validation of long noncoding RNA biomarkers in human nonsmall-cell lung carcinomas" 10 : 645-654, 2015

15 Meinshausen N, "High-dimensional graphs and variable selection with the lasso" 34 : 1436-1462, 2006

16 Jordan MI, "Graphical models: foundations of neural computation. Computational neuroscience series" The MIT Press 2001

17 Ali A, "Generalized pseudolikelihood methods for inverse covariance estimation" 54 : 280-288, 2017

18 Menéndez P, "Gene regulatory networks from multifactorial perturbations using graphical Lasso: application to the DREAM4 challenge" 5 : 14147-, 2010

19 Coloigner J, "Functional connectivity analysis for thalassemia disease based on a graphical lasso model" 2016

20 Chen J, "Extended Bayesian information criteria for model selection with large model spaces" 95 : 759-771, 2008

21 Foygel R, "Extended Bayesian information criteria for Gaussian graphical models" 2010

22 Mouallif M, "Expression profile of undifferentiated cell transcription factor 1 in normal and cancerous human epithelia" 95 : 251-259, 2014

23 Sun YB, "Expression of KISS1 and KISS1R (GPR54) may be used as favorable prognostic markers for patients with nonsmall cell lung cancer" 43 : 521-530, 2013

24 Pounds S, "Estimating the occurrence of false positives and false negatives in microarray studies by approximating and partitioning the empirical distribution of p-values" 19 : 1236-1242, 2003

25 Cai T, "Estimating sparse precision matrix: optimal rates of convergence and adaptive estimation" 44 : 455-488, 2016

26 Guo J, "Estimating heterogeneous graphical models for discrete data with an application to roll call voting" 9 : 821-848, 2015

27 Barabási A, "Emergence of scaling in random networks" 286 : 509-512, 1999

28 Benjamini Y, "Controlling the False Discovery Rate:A practical and powerful approach to multiple testing" 57 : 289-300, 1995

29 Wang T, "Consistent tuning parameter selection in high dimensional sparse linear regression" 102 : 1141-1151, 2011

30 Jemal A, "Cancer statistics" 60 : 277-300, 2010

31 Li X, "Biomarkers in the lung cancer diagnosis: a clinical perspective" 59 : 500-507, 2012

32 Khare K, "A convex pseudolikelihood framework for high dimensional partial correlation estimation with convergence guarantees" 77 : 803-825, 2015

33 Cai T, "A constrained l (1) minimization approach to sparse precision matrix estimation" 106 : 594-607, 2011

34 Bolstad BM, "A comparison of normalization methods for high density oligonucleotide array data based on variance and bias" 19 : 185-193, 2003

35 Tang H, "A 12-gene set predicts survival benefits from adjuvant chemotherapy in non-small cell lung cancer patients" 19 : 1577-1586, 2013