Synchronization over $Z_2$ and community detection in multiplex signed networks with constraints

TL;DR

This paper addresses the problem of community detection in multiplex signed networks using synchronization over
7 extsubscript{2} with constraints, comparing algorithms and analyzing robustness to noise, with applications to political voting data.

Contribution

It introduces and compares spectral, SDP, and message passing algorithms for
7 extsubscript{2} synchronization with constraints, analyzing their robustness and applicability to real-world data.

Findings

Message passing outperforms eigenvector method in certain conditions.

Robustness of eigenvector method analyzed using random matrix theory.

Algorithms successfully identify political communities in U.S. Congress data.

Abstract

Finding group elements from noisy measurements of their pairwise ratios is also known as the group synchronization problem, first introduced in the context of the group SO(2) of planar rotations, whose usefulness has been demonstrated recently in engineering and structural biology. Here, we focus on synchronization over $\mathbb{Z}_2$, and consider the problem of identifying communities in a multiplex network when the interaction between the nodes is described by a signed (possibly weighted) measure of similarity, and when the multiplex network has a natural partition into two communities. When one has the additional information that certain subsets of nodes represent the same unknown group element, we consider and compare several algorithms based on spectral, semidefinite programming (SDP) and message passing algorithms. In other words, all nodes within such a subset represent the same unknown group element, and one has available noisy pairwise measurements between pairs of nodes that belong to different non-overlapping subsets. Following a recent analysis of the eigenvector method for synchronization over SO(2), we analyze the robustness to noise of the eigenvector method for synchronization over $\mathbb{Z}_2$, when the underlying graph of pairwise measurements is the Erd\H{o}s-R\'{e}nyi random graph, using results from the random matrix theory literature. We also propose a message passing synchronization algorithm that outperforms the existing eigenvector synchronization algorithm only for certain classes of graphs and noise models, and enjoys the flexibility of incorporating constraints that may not be easily accommodated by other spectral or SDP-based methods. We apply our methods both to several synthetic models and a real data set of roll call voting patterns in the U.S. Congress across time, to identify the two existing communities, i.e., the Democratic and Republican parties.