Relational Information Bottleneck
This site contains material supplementary to the paper “Applying the information bottleneck to statistical relational learning” by Fabrizio Riguzzi and Nicola Di Mauro.
Attached to this page you can find a technical report detailing the derivation of the formulas used in the code.
The algorithm RIB and EM are distributed with YAP in the folder packages/cplint/rib and em.
To execute RIB or EM, prepare three files in the folder:
- stem.kb: contains the example interpretations
- stem.cpl: contains the LPAD for you which you want to learn the parameters
- stem.l: contains language bias information
stem can be replaced by any other name. Examples of these files can be found in the dataset pages.
To execute RIB, load rib.pl and call:
To execute EM, load em.pl and call:
More details can be found in the cplint manual (html, pdf).
Supplementary material can be downloaded here.
Fabrizio Riguzzi and Nicola Di Mauro. Applying the information bottleneck to statistical relational learning. Machine Learning, 86(1):89-114, 2012. The original publication is available at http://www.springerlink.com. [ bib | DOI | .pdf ]