Distributed principal component analysis on networks via directed graphical models

Author

Meng, Zhaoshi ; Wiesel, Ami ; Hero, Alfred O., III

Author_Institution

Univ. of Michigan, Ann Arbor, MI, USA

fYear

2012

fDate

25-30 March 2012

Firstpage

2877

Lastpage

2880

Abstract

We introduce an efficient algorithm for performing distributed principal component analysis (PCA) on directed Gaussian graphical models. By exploiting structured sparsity in the Cholesky factor of the inverse covariance (concentration) matrix, our proposed DDPCA algorithm computes global principal subspace estimation through local computation and message passing. We show significant performance and computation/communication advantages of DDPCA for online principal subspace estimation and distributed anomaly detection in real-world computer networks.

Keywords

Gaussian processes; computer network security; covariance matrices; graph theory; message passing; network theory (graphs); principal component analysis; Cholesky factor; DDPCA algorithm; computer networks; concentration matrix; directed Gaussian graphical models; distributed anomaly detection; distributed principal component analysis; global principal subspace estimation; inverse covariance matrix; message passing; online principal subspace estimation; structured sparsity; Computational modeling; Covariance matrix; Estimation; Graphical models; Matrix decomposition; Principal component analysis; Vectors; Graphical models; anomaly detection; distributed PCA; principal component analysis; subspace tracking;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on

Conference_Location

Kyoto

ISSN

1520-6149

Print_ISBN

978-1-4673-0045-2

Electronic_ISBN

1520-6149

Type

conf

DOI

10.1109/ICASSP.2012.6288518

Filename

6288518