Represents sufficient statistics produced by E-step of EM algorithm.
Learns parameters of Bayesian Network with Expectation Maximisation algorithm.
Learns parameters of Bayesian Network with Expectation Maximisation algorithm.
Cluster graph, which parameters are learned for
EM learning expects this cluster graph to contain initial cluster potentials in a form of table CPTs. For instance, for a cluster initial potentials: Factor(winterVar, sprinklerVar, Array(0.6, 0.4, 0.55, 0.45)), all variables except the last one (sprinkler) act as conditioning variables.
At the end of learning process, cluster graph is updated with the latest learned cluster potentials. While learning parameters, cluster graph is used for performing inference during expectation step of EM algorithm.
To inform EM algorithm about shared cluster initial potentials, for instance while learning unrolled Dynamic Bayesian Network, use clusterTypeId field on a Cluster object.
Data set used for learning parameters of Bayesian Network
Maximum number of iterations for which EM algorithm is executed
Progress monitoring. It is called by this method at the end of every iteration
Default implementation of EM algorithm.