Geodesic
An alternating minimization algorithm for Factor Analysis
Kybernetika, Tome 55 (2019) no. 4, pp. 740-754
Cet article a éte moissonné depuis la source Czech Digital Mathematics Library

Voir la notice de l'article

The problem of decomposing a given covariance matrix as the sum of a positive semi-definite matrix of given rank and a positive semi-definite diagonal matrix, is considered. We present a projection-type algorithm to address this problem. This algorithm appears to perform extremely well and is extremely fast even when the given covariance matrix has a very large dimension. The effectiveness of the algorithm is assessed through simulation studies and by applications to three real benchmark datasets that are considered. A local convergence analysis of the algorithm is also presented.
The problem of decomposing a given covariance matrix as the sum of a positive semi-definite matrix of given rank and a positive semi-definite diagonal matrix, is considered. We present a projection-type algorithm to address this problem. This algorithm appears to perform extremely well and is extremely fast even when the given covariance matrix has a very large dimension. The effectiveness of the algorithm is assessed through simulation studies and by applications to three real benchmark datasets that are considered. A local convergence analysis of the algorithm is also presented.
DOI : 10.14736/kyb-2019-4-0740
Classification : 62H25
Keywords: matrix decomposition; factor analysis; covariance matrices; low rank matrices; projections 
@article{10_14736_kyb_2019_4_0740,
     author = {Ciccone, Valentina and Ferrante, Augusto and Zorzi, Mattia},
     title = {An alternating minimization algorithm for {Factor} {Analysis}},
     journal = {Kybernetika},
     pages = {740--754},
     year = {2019},
     volume = {55},
     number = {4},
     doi = {10.14736/kyb-2019-4-0740},
     mrnumber = {4043546},
     zbl = {07177914},
     language = {en},
     url = {http://geodesic.mathdoc.fr/articles/10.14736/kyb-2019-4-0740/}
}
TY  - JOUR
AU  - Ciccone, Valentina
AU  - Ferrante, Augusto
AU  - Zorzi, Mattia
TI  - An alternating minimization algorithm for Factor Analysis
JO  - Kybernetika
PY  - 2019
SP  - 740
EP  - 754
VL  - 55
IS  - 4
UR  - http://geodesic.mathdoc.fr/articles/10.14736/kyb-2019-4-0740/
DO  - 10.14736/kyb-2019-4-0740
LA  - en
ID  - 10_14736_kyb_2019_4_0740
ER  - 
%0 Journal Article
%A Ciccone, Valentina
%A Ferrante, Augusto
%A Zorzi, Mattia
%T An alternating minimization algorithm for Factor Analysis
%J Kybernetika
%D 2019
%P 740-754
%V 55
%N 4
%U http://geodesic.mathdoc.fr/articles/10.14736/kyb-2019-4-0740/
%R 10.14736/kyb-2019-4-0740
%G en
%F 10_14736_kyb_2019_4_0740
Ciccone, Valentina; Ferrante, Augusto; Zorzi, Mattia. An alternating minimization algorithm for Factor Analysis. Kybernetika, Tome 55 (2019) no. 4, pp. 740-754. doi: 10.14736/kyb-2019-4-0740

[1] Agarwal, A., Negahban, S., Wainwright, M. J.: Noisy matrix decomposition via convex relaxation: Optimal rates in high dimensions. Ann. Statist. 40 (2012), 2, 1171-1197. | DOI | MR

[2] Bai, J., Li, K., al., et: Statistical analysis of factor models of high dimension. Ann. Statist. 40 (2012), 1, 436-465. | DOI | MR

[3] Bai, J., Ng, S.: Determining the number of factors in approximate factor models. Econometrica 70 (2002), 1, 191-221. | DOI | MR

[4] Bai, J., Ng, S.: Large dimensional factor analysis. Found. Trends Econometr. 3 (2008), 2, 89-163. | DOI

[5] Bertsimas, D., Copenhaver, M. S., Mazumder, R.: Certifiably optimal low rank factor analysis. J. Machine Learning Res. 18 (2017), 29, 1-53. | MR

[6] Ciccone, V., Ferrante, A., Zorzi, M.: Factor analysis with finite data. In: Proc. 56th Annual Conference on Decision and Control (CDC), VIC, Melbourne 2017, pp. 4046-4051. | DOI

[7] Ciccone, V., Ferrante, A., Zorzi, M.: Robust identification of "Sparse Plus Low-rank" graphical models: An optimization approach. In: Proc. 2018 IEEE Conference on Decision and Control (CDC), Miami Beach 2018, pp. 2241-2246. | DOI

[8] Ciccone, V., Ferrante, A., Zorzi, M.: Factor models with real data: A robust estimation of the number of factors. IEEE Trans. Automat. Control 64 (2019), 6, 2412-2425. | DOI | MR

[9] Deistler, M., Zinner, C.: Modelling high-dimensional time series by generalized linear dynamic factor models: An introductory survey. Comm. Inform. Systems 7 (2007), 2, 153-166. | DOI | MR

[10] Riccia, G. Della, Shapiro, A.: Minimum rank and minimum trace of covariance matrices. Psychometrika 47 (1982), 443-448. | DOI | MR

[11] Fazel, M.: Matrix rank minimization with applications. Elec. Eng. Dept. Stanford University 54 (2002), 1-130.

[12] Fazel, M., Hindi, H., Boyd, S.: Rank minimization and applications in system theory. In: Proc. American Control Conference 4 (2004), pp. 3273-3278. | DOI

[13] Geweke, J.: The dynamic factor analysis of economic time series models. In: Latent Variables in Socio-Economic Models, SSRI workshop series, North-Holland 1977, pp. 365-383.

[14] Guttman, L.: Some necessary conditions for common-factor analysis. Psychometrika 19 (1954), 2, 149-161. | DOI | MR

[15] Harman, H. H., Jones, W. H.: Factor analysis by minimizing residuals (minres). Psychometrika 31 (1066), 3, 351-368. | DOI | MR

[16] Heij, C., Scherrer, W., Deistler, M.: System identification by dynamic factor models. SIAM J. Control Optim. 35 (1997), 6, 1924-1951. | DOI | MR

[17] Helmke, U., Shayman, M. A.: Critical points of matrix least squares distance functions. Linear Algebra Appl. 215 (1995), 1-19. | DOI | MR

[18] Lax, P. D.: Linear Algebra and Its Applications. Second edition. Wiley-Interscience, 2007. | MR

[19] Lewis, A. S., Malick, J.: Alternating projections on manifolds. Math. Oper. Res. 33 (2008), 1, 216-234. | DOI | MR

[20] Ning, L., Georgiou, T. T., Tannenbaum, A., Boyd, S. P.: Linear models based on noisy data and the Frisch scheme. SIAM Rev. 57 (2015), 2, 167-197. | DOI | MR

[21] Reiersøl, O.: Identifiability of a linear relation between variables which are subject to error. Econometrica: J. Economet. Soc. 18 (1950), 4, 375-389. | DOI | MR

[22] Scherrer, W., Deistler, M.: A structure theory for linear dynamic errors-in-variables models. SIAM J. Control Optim. 36 (1998), 6, 2148-2175. | DOI | MR

[23] Shapiro, A.: Rank-reducibility of a symmetric matrix and sampling theory of minimum trace factor analysis. Psychometrika 47 (1982), 2, 187-199. | DOI | MR

[24] Shapiro, A., Berge, J. M. F. Ten: Statistical inference of minimum rank factor analysis. Psychometrika 67 (2002), 1, 79-94. | DOI | MR

[25] Zorzi, M., Sepulchre, R.: AR identification of latent-variable graphical models. IEEE Trans. Automat. Control 61 (2016), 9, 2327-2340. | DOI | MR

Cité par Sources :