On the complexity of dynamic directed evidential networks with conditional belief functions construction and belief propagation
Résumé
Directed evidential networks obtain their efficiency by representing (in)dependencies between variables in the model. Based on the framework of evidence theory which is known as a general framework for representing knowledge and reasoning under uncertainty, directed evidential networks are a generalization of Bayesian networks. This paper presents a new dynamic evidential model for representing uncertainty and managing temporal changes in data. This proposed model offers an alternative framework for dynamic probabilistic and dynamic possibilistic networks. A complexity study of the representation and the reasoning in the proposed model is also presented in this paper.