Bayesian Networks for Ruby ( bn4r )

bn4r is a bayesian networks library on ruby that provides the user with classes for create bayesian networks and diverse algorithms for solve them.

Its algorithms implementation are based on: S.Russell, P.Norving, "Artificial Intelligence, A Modern Approach", 2nd Edition.

Website: http://bn4r.rubyforge.org

Spanish Website: http://bn4r.rubyforge.org/es

Rubyforge Project: http://rubyforge.org/projects/bn4r

Dependencies

Design principles

The library consists on the object BayesNet thinked to be filled with BayesNetNode, these objects are defined in bn.rb. BayesNet object is a especialization of RGL::DirectedAdjacencyGraph ( http://rgl.rubyforge.org ).

The file bn_algorithms.rb has the implementation of the inference algorithms that can be used to solve BayesNet structures.

Files bn_export.rb and bn_import.rb have methods for import and export bayesian networkd in different formats.

Finally, a set of objects and methods are given to automaticly fill BayesNetNode probabilities tables.

Play a bit ...

  1. Install the gem ( bn4r-0.1.2.gem ) gem install bn4r

  2. Include the bn4r require 'bn4r'

  3. Create your first bayes net

    Create BayesNet

    bn_aima = BayesNet.new

    Create nodes for the Bayes Net (BayesNetNodes)

    burglary = BayesNetNode.new("Burglary") earthquake = BayesNetNode.new("Earthquake") alarm = BayesNetNode.new("Alarm") john_calls = BayesNetNode.new("JohnCalls") mary_calls = BayesNetNode.new("MaryCalls")

    Add nodes ( vertex ) to the BayesNet

    bn_aima.add_vertex(burglary) bn_aima.add_vertex(earthquake) bn_aima.add_vertex(alarm) bn_aima.add_vertex(john_calls) bn_aima.add_vertex(mary_calls)

    Add relations ( edges ) between nodes in the BayesNet

    bn_aima.add_edge(burglary,alarm) bn_aima.add_edge(earthquake,alarm) bn_aima.add_edge(alarm,john_calls) bn_aima.add_edge(alarm,mary_calls)

    Assign probabilities to each node

    burglary.set_probability_table([], [0.001, 0.999] ) earthquake.set_probability_table([], [0.002, 0.998] )

    alarm.set_probability_table([burglary,earthquake], [0.95, 0.05, 0.94, 0.06, 0.29, 0.61, 0.001,0.999] )

    john_calls.set_probability_table([alarm], [0.90,0.10,0.05,0.95]) mary_calls.set_probability_table([alarm], [0.70,0.30,0.01,0.99])

  4. Solve it!

    John and Mary are calling ...

    john_calls.set_value(true)
    mary_calls.set_value(true)

    Why?

    is_there_a_burglary = bn_aima.enumeration_ask( burglary, [john_calls, mary_calls] ) puts "Call the police!" if is_there_a_burglary > is_there_a_burglary

    is_the_alarm_on = bn_aima.enumeration_ask( alarm, [john_calls, mary_calls] ) puts "Run home, your alarm is distubing the neigborhood!" if is_the_alarm_on > is_the_alarm_on

    is_there_a_earthquake = bn_aima.enumeration_ask( earthquake, [john_calls, mary_calls] ) puts "Calm yourself, there isn't a earthquake ;)" if is_there_a_earthquake < is_there_a_earthquake

  5. See how your bayes net looks like

    In .dot format

    bn_aima.to_dot

    In Microsoft Belief Networks (.xbn) format

    (download for free in: http://research.microsoft.com/adapt/MSBNx )

    bn_aima.to_xbn

Documentation

Documentation can be found at http://bn4r.rubyforge.org/rdoc or can be generated using rdoc tool under the source code with:

rdoc README lib

Credits

Thanks to N?ria Bel ( http://www.upf.edu/pdi/iula/nuria.bel ) for her work in this project without her it cannot be done.

Thanks to Ryan Dahl for his work in http://www.math.rochester.edu/people/grads/rld/bayesnets that was the inspiration of the project.

Also thanks to all the ruby community.

Copying

This work is developed by Sergio Espeja ( http://www.upf.edu/pdi/iula/sergio.espeja, sergio.espeja at gmail.com ) mainly in Institut Universitari de Ling??stica Aplicada of Universitat Pompeu Fabra ( http://www.iula.upf.es ), and also in bee.com.es ( http://bee.com.es ).

It is free software, and may be redistributed under GPL license.

Support

Please contact me in http://rubyforge.org/projects/bn4r.