Scalable MCMC for Mixed Membership Stochastic Blockmodels

Abstract

We propose a stochastic gradient Markov chain Monte Carlo (SG-MCMC) algorithm for scalable inference in mixed-membership stochastic blockmodels (MMSB). Our algorithm is based on the stochastic gradient Riemannian Langevin sampler and achieves both faster speed and higher accuracy at every iteration than the current state-of-the-art algorithm based on stochastic variational inference. In addition we develop an approximation that can handle models that entertain a very large number of communities. The experimental results show that SG-MCMC strictly dominates competing algorithms in all cases.

9 pages, 18 figures

References in corpus (1)

• Bayesian Posterior Sampling via Stochastic Gradient Fisher Scoring

Cited by in corpus (4)

• Bayesian Graph Convolutional Neural Networks using Node Copying
• Learning a Latent Simplex in Input-Sparsity Time
• sgmcmc: An R Package for Stochastic Gradient Markov Chain Monte Carlo
• Small-Variance Asymptotics for Nonparametric Bayesian Overlapping Stochastic Blockmodels