EM and SEM Learning now available
An EM Learning algorithm (incomplete data) has been implemented as well as a S-EM Algorithm (Structure Learning with incomplete data)
Francois de Brouchoven has just contributed to OpenBayes by implementing the following features :
EM Learning
Structure Learning
This allows one to learn the structure of a network (the edges that link the variables) by simply providing a set of stochastic variables (with or without any edges) and a set of complete data.
S-EM Learning
Also learns the structure of a BN, but does not need a set of complete data. Any data missing is estimated using the EM algorithm above
Thank you François for this wonderful contribution.