Bayesian network Learning and Inference Project

This package implements the algorithms detailed in the following papers:

• Efficient learning of bounded-treewidth Bayesian networks from complete and incomplete data sets (IJAR 2018) - supplementary material
• Approximated Structural Learning for Large Bayesian Networks (accepted for publication ML 2018) - supplementary material
• Learning Treewidth-Bounded Bayesian Networks with Thousands of Variables (NIPS 2016) Mauro Scanagatta, Giorgio Corani, Cassio P. de Campos, Marco Zaffalon

The process of learning a bounded-treewidth BN is explained by using the "child" network as example.

The format for the initial dataset has to be the same as the file "child-5000.dat", namely a space-separated file containing:

* First line: list of variables names, separated by space;
* Second line: list of variables cardinalities, separated by space;
* Following lines: list of values taken by the variables in each datapoint, separated by space.

• -d VAL : Datafile path (.dat format)
• -j VAL : Parent set scores output file (.jkl format)
• -r VAL : Structure output file (.res format)
• -t N : Maximum time limit, in seconds (default: 10)
• -b N : Number of machine cores to use (default: 1)
• -w N : Maximum treewidth
• -seed N : Seed for the pseudo random number generator

The first step is build the parent sets score cache. It can be done with:

java -jar blip.jar scorer.sq -c bdeu -d data/child-5000.dat -j data/child-5000.jkl -n 3 -t 10

• -a N : (if BDeu is chosen) equivalent sample size parameter (default: 1.0)
• -c VAL : Chosen score function. Possible choices: BIC, BDeu (default: bic)
• -n N : Maximum learned in-degree (if 0, no constraint is applied) (default: 0)

For perfoming with k-greedy:

java -jar blip.jar solver.kg -j data/child-5000.jkl -r data/child.kg.res -t 10 -w 4 -v 1

For perfoming with the k-greedy enhanched by entropy-based sample ordering:

java -jar blip.jar solver.kg.adv -smp ent -d data/child-5000.dat -j data/child-5000.jkl -r data/child-5000.kgent.res -t 10 -w 4 -v 1

For perfoming with k-max:

java -jar blip.jar solver.kmax -j data/child-5000.jkl -r data/child-5000.kmax.res -t 10 -w 4 -v 1

The format of the ".res" file is as follows: each line indicates the parent set assigned to each variable and its score.

For example the line "4: -2797.39 (10,17,18)" indicates that to the variable with index 4 in the dataset are assgined as parents the variables with index (10,17,18). This parent set has score -2797.39 (by default the score function is the BIC).

Using the structure found it is possible to learn the parameters with:

java -jar blip.jar parle -d data/child-5000.dat -r data/child-5000.kmax.res -n data/child-5000.kmax.uai

The final output will be a full Bayesian network in UAI format.