Scalable MCMC for Mixed Membership Stochastic Blockmodels
This work addresses the problem of efficient inference for large-scale network models, offering a scalable solution for researchers and practitioners, though it appears incremental as it builds on existing MCMC and stochastic gradient techniques.
The authors tackled scalable inference in mixed-membership stochastic blockmodels by proposing a stochastic gradient MCMC algorithm, which achieved faster speed and higher accuracy than state-of-the-art methods in all experimental cases.
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.