Probabilistic Graphical Model Toolbox

Mens X Machina Probabilistic Graphical Model Toolbox (MxM PGM) aims to provide a comprehensive set of tools for Bayesian networks and other probabilistic graphical models. Currently only local Bayesian network learning and Bayesian network skeleton identification using the MMPC algorithm is implemented. Support for full structure learning and inference will be added in future versions.

Mens X Machina Probabilistic Graphical Model Samples (MxM PGM Samples) is a collection of samples from several Bayesian networks. MxM PGM Samples is a supplement to MxM PGM.

MxM PGM and its supplement MxM PGM Samples are released under GPLv3.

Demos

• Creating, viewing and sampling a Bayesian network
• Learning the parents and children of a Bayesian network node
• Identifying the skeleton of a Bayesian network

The demos are also included in the toolbox download.

Requirements

• MATLAB 7.9 (R2009b) or later
• MATLAB Statistics Toolbox ™
• MATLAB Bioinformatics Toolbox ™
• MATLAB Neural Network Toolbox ™ (for some functions)
• Mens X Machina Commons Toolbox 0.9.2
• MATLAB xUnit Test Framework in order to run the unit sets

Download

• Mens X Machina Probabilistic Graphical Model Toolbox 0.9.2 (.zip file, 1.5 MB)
• Mens X Machina Probabilistic Graphical Model Samples 1.0 (.zip file, 64.3 MB)

Installation

To use MxM PGM in MATLAB, add the "pgm/pgm" directory to the MATLAB path. See the MATLAB documentation for setting the search path. Mens X Machina Commons Toolbox must be also on the MATLAB path. The “pgm/tests” directory contains MATLAB xUnit tests. xUnit must be on the MATLAB path in order to run the tests.

Bibliography

• The max-min hill-climbing Bayesian network structure learning algorithm
• Local Causal and Markov Blanket Induction for Causal Discovery and Feature Selection for Classification Part I: Algorithms and Empirical Evaluation

Previous versions

• Mens X Machina Probabilistic Graphical Model Toolbox 0.9.1 (.zip file, 2.1 MB)
• Mens X Machina Bayesian Network Toolbox 0.9 (.zip file, 2 MB)

Contact

For any question, suggestion or bug report please contact Angelos P. Armen.