MPF is a technique for parameter estimation in un-normalized probabilistic models. It is described in the paper:
J Sohl-Dickstein, P Battaglino, MR DeWeese
Minimum probability flow learning
International Conference on Machine Learning (2011)
http://arxiv.org/abs/0906.4779
This repository builds on the work done by Sohl-Dickstein, combining the use of MPF with the use of an auxiliary Markov Random Field in order to make edge weight estimation embarrassingly parallel.