Class BoaEbnaModel

java.lang.Object
com.knezevic.edaf.v3.models.discrete.BoaEbnaModel
All Implemented Interfaces:
Model<BitString>

public final class BoaEbnaModel extends Object implements Model<BitString>
Sparse Bayesian-network model used by BOA/EBNA-style discrete EDAs.

Variables are ordered by entropy and each node chooses up to maxParents predecessors with highest mutual information. The resulting directed acyclic graph defines:

   p(x) = Π_i p(x_i | Pa(x_i))
 
with Laplace-smoothed conditional probability tables.

Compared to tree models, this representation captures higher-order local structure while keeping sampling exact and deterministic given RNG stream.