Column generation for exact BN learning: Work in progress
James Cussens · 2012
In existing integer linear programming (ILP) approaches to learning the structure of a Bayesian network (BN) from data ([1, 4, 2]) a bi-nary variable I(W → u) is created for each BN variable u ∈ V and each candidate parent set W. I(W → u) takes the value 1 iff W is