Provides {@link de.jstacs.sequenceScores.statisticalModels.differentiable.DifferentiableStatisticalModel}s that are directed graphical models. The currently implemented
{@link de.jstacs.sequenceScores.statisticalModels.differentiable.directedGraphicalModels.BayesianNetworkDiffSM} provides structure learning for inhomogeneous Markov models, Bayesian trees, and permuted Markov models,
but can be extended to other graph structures easily using the {@link de.jstacs.sequenceScores.statisticalModels.differentiable.directedGraphicalModels.structureLearning.measures.Measure} interface.
It is important to define only moral Bayesian network structures because of the chosen parameterization.