A Riemannian Optimization Approach for Role Model Extraction

The ability to compute meaningful clusters of nodes is important in the analysis of large networks. A particular approach to this problem is the use of role models of a graph. For large networks, the algorithms must be specifically designed to extract role models while maintaining efficiency in storage and computations. Browet et al. have investigated the computation of role models for both moderately sized and large networks. They proposed an efficient iteration on lowrank matrices to compute an approximation to the required pairwise node similarity measure at MTNS 2014. In this paper, we summarize a new approach to compute an approximation to the pairwise node similarity measure for large networks based on Riemannian optimization. A comparison of our optimization approach with that of Browet et al. shows that our approach computes the same approximate solution in significantly less time.

[1]  Levent Tunçel,et al.  Optimization algorithms on matrix manifolds , 2009, Math. Comput..

[2]  Wen Huang,et al.  A Broyden Class of Quasi-Newton Methods for Riemannian Optimization , 2015, SIAM J. Optim..

[3]  Francis R. Bach,et al.  Low-Rank Optimization on the Cone of Positive Semidefinite Matrices , 2008, SIAM J. Optim..

[4]  Wen Huang,et al.  Optimization algorithms on Riemannian manifolds with applications , 2013 .

[5]  Paul Van Dooren,et al.  Rank-constrained optimization: a Riemannian manifold approach , 2015, ESANN.

[6]  Wen Huang,et al.  A Riemannian symmetric rank-one trust-region method , 2014, Mathematical Programming.

[7]  Santo Fortunato,et al.  Community detection in graphs , 2009, ArXiv.

[8]  Arnaud Browet,et al.  Fast community detection using local neighbourhood search , 2013, ArXiv.

[9]  C. Baker.  RIEMANNIAN MANIFOLD TRUST-REGION METHODS WITH APPLICATIONS TO EIGENPROBLEMS , 2008 .

[10]  Vincent A. Traag,et al.  Algorithms and Dynamical Models for Communities and Reputation in Social Networks , 2014 .

[11]  Paul Van Dooren,et al.  A MEASURE OF SIMILARITY BETWEEN GRAPH VERTICES . WITH APPLICATIONS TO SYNONYM EXTRACTION AND WEB SEARCHING , 2002 .

[12]  Sabine Van Huffel,et al.  A Geometric Newton Method for Oja's Vector Field , 2008, Neural Computation.

[13]  Thomas P. Cason,et al.  Role extraction in networks , 2012 .

[14]  Mason A. Porter,et al.  Communities in Networks , 2009, ArXiv.

[15]  Arnaud Browet,et al.  Low-rank Similarity Measure for Role Model Extraction , 2013, ArXiv.

[16]  Stanley Wasserman,et al.  Social Network Analysis: Methods and Applications , 1994, Structural analysis in the social sciences.

[17]  Arnaud Browet,et al.  Algorithms for community and role detection in networks , 2014 .