Consistent Estimation of Mixed Memberships with Successive Projections

TL;DR

This paper introduces SPOC, a new algorithm for estimating mixed memberships in community detection within MMSB, combining spectral clustering and geometric methods, and proves its consistency with strong experimental performance.

Contribution

The paper presents SPOC, a novel algorithm that integrates spectral clustering with geometric approaches for consistent mixed membership estimation in MMSB.

Findings

SPOC is provably consistent under general MMSB conditions.

SPOC outperforms existing algorithms in experiments.

The method effectively handles overlapping community structures.

Abstract

This paper considers the parameter estimation problem in Mixed Membership Stochastic Block Model (MMSB), which is a quite general instance of random graph model allowing for overlapping community structure. We present the new algorithm successive projection overlapping clustering (SPOC) which combines the ideas of spectral clustering and geometric approach for separable non-negative matrix factorization. The proposed algorithm is provably consistent under MMSB with general conditions on the parameters of the model. SPOC is also shown to perform well experimentally in comparison to other algorithms.