English

A Theoretical Case Study of Structured Variational Inference for Community Detection

Statistics Theory 2020-03-03 v5 Statistics Theory

Abstract

Mean-field variational inference (MFVI) has been widely applied in large scale Bayesian inference. However MFVI, which assumes a product distribution on the latent variables, often leads to objective functions with many local optima, making optimization algorithms sensitive to initialization. In this paper, we study the advantage of structured variational inference for the two class Stochastic Blockmodel. The variational distribution is constructed to have pairwise dependency structure on the nodes of the network. We prove that, in a broad density regime and for general random initializations, unlike MFVI, the class labels estimated from our method converge to the ground truth with high probability, when the model parameters are known, estimated within a reasonable range or jointly optimized with the variational parameters. In addition, empirically we demonstrate structured VI is more robust compared with MFVI when the graph is sparse and the signal to noise ratio is low. The paper takes a first step towards understanding the importance of dependency structure in variational inference for community detection.

Keywords

Cite

@article{arxiv.1907.12203,
  title  = {A Theoretical Case Study of Structured Variational Inference for Community Detection},
  author = {Mingzhang Yin and Y. X. Rachel Wang and Purnamrita Sarkar},
  journal= {arXiv preprint arXiv:1907.12203},
  year   = {2020}
}

Comments

AISTATS 2020

R2 v1 2026-06-23T10:33:21.873Z