English

Robustness of spectral methods for community detection

Data Structures and Algorithms 2019-06-27 v2 Probability

Abstract

The present work is concerned with community detection. Specifically, we consider a random graph drawn according to the stochastic block model~: its vertex set is partitioned into blocks, or communities, and edges are placed randomly and independently of each other with probability depending only on the communities of their two endpoints. In this context, our aim is to recover the community labels better than by random guess, based only on the observation of the graph. In the sparse case, where edge probabilities are in O(1/n)O(1/n), we introduce a new spectral method based on the distance matrix D(l)D^{(l)}, where Dij(l)=1D^{(l)}_{ij} = 1 iff the graph distance between ii and jj, noted d(i,j)d(i, j) is equal to \ell. We show that when clog(n)\ell \sim c\log(n) for carefully chosen cc, the eigenvectors associated to the largest eigenvalues of D(l)D^{(l)} provide enough information to perform non-trivial community recovery with high probability, provided we are above the so-called Kesten-Stigum threshold. This yields an efficient algorithm for community detection, since computation of the matrix D(l)D^{(l)} can be done in O(n1+κ)O(n^{1+\kappa}) operations for a small constant κ\kappa. We then study the sensitivity of the eigendecomposition of D(l)D^{(l)} when we allow an adversarial perturbation of the edges of GG. We show that when the considered perturbation does not affect more than O(nε)O(n^\varepsilon) vertices for some small ε>0\varepsilon > 0, the highest eigenvalues and their corresponding eigenvectors incur negligible perturbations, which allows us to still perform efficient recovery.

Keywords

Cite

@article{arxiv.1811.05808,
  title  = {Robustness of spectral methods for community detection},
  author = {Ludovic Stephan and Laurent Massoulié},
  journal= {arXiv preprint arXiv:1811.05808},
  year   = {2019}
}
R2 v1 2026-06-23T05:15:18.878Z