GRaphical Models in Mallet
From Mallet
GRMM is a toolkit for performing inference and learning in graphical models of arbitrary structure. Its main features are:
- Supports arbitrary factor graphs, which subsume both Markov random fields and Bayesian networks.
- Efficient implementations of several inference algorithms, including junction tree, belief propagation, and Gibbs sampling. All inference algorithms work for factors of any size (not just pairwise).
- Ability to train conditional random fields of arbitrary structure and parameter tying.
It is implemented as an add-on package to Mallet. Its name stands for GRaphical Models in Mallet. Only discrete variables are supported. It has been developed for several years, and results from GRMM have been used in several papers. It is written by Charles Sutton.
Obtaining GRMM
Download the GRMM source (http://mallet.cs.umass.edu/mallet/dist/grmm/grmm-0.1.2.tgz). The current version is 0.1.2. I have recently done a lot of renaming, so even though I have pretty good unit tests, some bugs may remain.
Installation instructions: Unpack and type make. Java 1.4 is required, 1.5 is fine.
Documentation
- If you just want to use the CRF implementation, here's a quick introduction to Command-Line CRFs in GRMM
- If you'd like to build your own factor graphs and call inference algorithms directly, then see Cooking with GRMM.
- Finally, Javadocs are available.
Citation
If you use GRMM to produce results in a publication, please cite it as
@unpublished{sutton06grmm,
author = "Charles Sutton",
title = "GRMM: A Graphical Models Toolkit",
note = "http://mallet.cs.umass.edu",
year = 2006}
This allows me to track the usage of GRMM.