This repository is associated with the article: Aliverti, E. and Russo, M. (2020). Stratified stochastic variational inference for high-dimensional network factor model [link].
In this article, we focus on the latent factor model for networks, a very general approach which encompasses the latent distance model and the stochastic block-model as special cases. We develop scalable algorithms to conduct approximate Bayesian inference via stochastic optimization. Leveraging sparse representations of network data, the proposed algorithms show massive computational and storage benefits, and allow to conduct inference in settings with thousands of nodes.
The repository contains the R package implementing svilf.
We recommend installing svilf locally in the lib/ folder, for example with
R CMD install -l lib ./svilf_0.1.tar.gz
The package provides wrappers around core c++ implementations, relying on Armdadillo's SpMat format.
Wrappers are written in R and provide easy tools to specify different options such as the maximum number of iterations, tolerance, starting values, step size sequence and others.
All functions provide default configurations.
The scripts in the folders APPLICATION and SIMULATIONS reproduce results from the paper, and are are set-up to use ./lib as the local library folder.