A Boolean Matrix Decomposition Framework
Boolean matrix decomposition is a method to obtain a compressed representation of a matrix with Boolean entries. BMaD is a modular framework that unifies several Boolean matrix decomposition algorithms, and provide methods to evaluate their performance. The main advantages of the framework are its modular approach and hence the flexible combination of the steps of a Boolean matrix decomposition and the capability of handling missing values. The framework is licensed under the GPLv3.
Generated Javadoc can be found at http://joergwicker.github.io/bmad/apidocs/.
We use maven as a build tool, so just run
mvn clean install
to build BMaD.
BMaD is in Maven Central, so you can just add this dependency to your pom.xml:
<dependency>
<groupId>nz.wicker</groupId>
<artifactId>bmad</artifactId>
<version>2.6</version>
</dependency>
The demo class can be executed using:
mvn exec:java -Dexec.mainClass="nz.wicker.bmad.demo.Demo"
If you want to cite BMaD in your publication, please cite the following ECML/PKDD paper:
Andrey Tyukin, Stefan Kramer, and Jörg Wicker.
"BMaD–A Boolean Matrix Decomposition Framework."
Machine Learning and Knowledge Discovery in Databases.
Springer Berlin Heidelberg, 2014. 481-484.
Bibtex entry:
@incollection{
year={2014},
isbn={978-3-662-44844-1},
booktitle={Machine Learning and Knowledge Discovery in Databases},
volume={8726},
series={Lecture Notes in Computer Science},
editor={Calders, Toon and Esposito, Floriana and H{\"u}llermeier, Eyke and Meo, Rosa},
doi={10.1007/978-3-662-44845-8_40},
title={BMaD – A Boolean Matrix Decomposition Framework},
url={http://dx.doi.org/10.1007/978-3-662-44845-8_40},
publisher={Springer Berlin Heidelberg},
author={Tyukin, Andrey and Kramer, Stefan and Wicker, J{\"o}rg},
pages={481-484},
language={English}
}