Skip to content
Discussion of Durante et al for JSM 2017. Includes factorial network model generalization.
Jupyter Notebook Python Shell
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Factorial Network Models

A discussion of "Nonparametric Bayes Modeling of Populations of Networks" by Durante, Dunson, and Vogelstein (2016).

Scott W. Linderman and David M. Blei


While the modeling of single networks has received much attention, we agree with Durante et al. that the modeling of populations of networks has been largely overlooked. Given that such data are increasingly common in fields such as neuroscience, their paper makes a timely contribution. In this discussion, we consider a factorial generalization of the proposed mixture of latent space models, and we suggest cases in which factorial models may naturally capture our intuition about the underlying generative process of the data. We compare these two models using the human brain data studied by Durante et al., and we suggest some avenues for future work.


  • data: helpers to download and extract the brain data.
  • doc: LaTeX source of the discussion paper.
  • lsm: source code for the latent space model (LSM), the mixture of latent space models (MoLSM), and the factorial latent space model (fLSM).
  • notebooks: analysis of synthetic data and brain data. This includes code to reproduce the figures in the paper.
  • results: directory where model fits will be cached.


First, make sure you have pypolyagamma installed. This can be done with pip or by source. See the link above for detailed installation instructions.


git clone
cd factorial-network-models
pip install -e .


Durante, Daniele, David B. Dunson, and Joshua T. Vogelstein. "Nonparametric Bayes modeling of populations of networks." Journal of the American Statistical Association (2016).

You can’t perform that action at this time.