Phase Transitions in Spectral Community Detection

被引:24
作者
Chen, Pin-Yu [1 ]
Hero, Alfred O., III [1 ]
机构
[1] Univ Michigan, Dept Elect Engn & Comp Sci, Ann Arbor, MI 48109 USA
关键词
Graph clustering; graph signal processing; network data analysis; spectral clustering; GRAPHS;
D O I
10.1109/TSP.2015.2442958
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Consider a network consisting of two subnetworks (communities) connected by some external edges. Given the network topology, the community detection problem can be cast as a graph partitioning problem that aims to identify the external edges as the graph cut that separates these two subnetworks. In this paper, we consider a general model where two arbitrarily connected subnetworks are connected by random external edges. Using random matrix theory and concentration inequalities, we show that when one performs community detection via spectral clustering there exists an abrupt phase transition as a function of the random external edge connection probability. Specifically, the community detection performance transitions from almost perfect detectability to low detectability near some critical value of the random external edge connection probability. We derive upper and lower bounds on the critical value and show that the bounds are equal to each other when two subnetwork sizes are identical. Using simulated and experimental data we show how these bounds can be empirically estimated to validate the detection reliability of any discovered communities.
引用
收藏
页码:4339 / 4347
页数:9
相关论文
共 43 条
[1]  
[Anonymous], 2013, Matrix analysis
[2]  
[Anonymous], 2011, PROC INT JOINT C ART
[3]   The singular values and vectors of low rank perturbations of large rectangular random matrices [J].
Benaych-Georges, Florent ;
Nadakuditi, Raj Rao .
JOURNAL OF MULTIVARIATE ANALYSIS, 2012, 111 :120-135
[4]   Seeing the Bigger Picture [J].
Bertrand, Alexander ;
Moonen, Marc .
IEEE SIGNAL PROCESSING MAGAZINE, 2013, 30 (03) :71-82
[5]   A nonparametric view of network models and Newman-Girvan and other modularities [J].
Bickel, Peter J. ;
Chen, Aiyou .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2009, 106 (50) :21068-21073
[6]  
Chen P.-Y., 2013, P IEEE GLOBALSIP
[7]   Universal phase transition in community detectability under a stochastic block model [J].
Chen, Pin-Yu ;
Hero, Alfred O., III .
PHYSICAL REVIEW E, 2015, 91 (03)
[8]   Assessing and Safeguarding Network Resilience to Nodal Attacks [J].
Chen, Pin-Yu ;
Hero, Alfred O., III .
IEEE COMMUNICATIONS MAGAZINE, 2014, 52 (11) :138-143
[9]  
Chung F., 1996, Spectral Graph Theory, DOI DOI 10.1090/CBMS/092
[10]   Inference and Phase Transitions in the Detection of Modules in Sparse Networks [J].
Decelle, Aurelien ;
Krzakala, Florent ;
Moore, Cristopher ;
Zdeborova, Lenka .
PHYSICAL REVIEW LETTERS, 2011, 107 (06)