Credal Compositional Models and Credal Networks
نویسنده
چکیده
This paper studies the composition operator for credal sets introduced at the last ISIPTA conference in more detail. Our main attention is devoted to the relationship between a special type of compositional model, so-called perfect sequences of credal sets, and those of (precise) probability distributions, with the goal of finding the relationship between credal compositional models and credal networks. We prove that a perfect sequence of credal sets is a convex hull of perfect sequences of extreme points of these credal sets. Finally, we reveal the relationship among credal networks (in a general sense), perfect sequences of credal sets and separately specified credal networks.
منابع مشابه
Credal Networks and Compositional Models: Preliminary Considerations
In the paper we present basic concepts concerning credal networks and compositional models for credal sets and describe the problem of imprecision increase in the first type of these models.
متن کاملOperator of Composition for Credal Sets
This paper is the first attempt to introduce the operator of composition, already known from probability, possibility and evidence theories, also for credal sets. We prove that the proposed definition preserves all the necessary properties of the operator enabling us to define compositional models as an efficient tool for multidimensional models representation. Theoretical results are accompani...
متن کاملSet-Based Variational Methods in Credal Networks: the SV2U Algorithm
Abstract. Graphical models that represent uncertainty through sets of probability measures are often referred to as credal networks. Polynomial-time exact inference methods are available only for polytree-structured binary credal networks. In this work, we approximate potentially intractable inferences in multiconnected binary networks by tractable inferences in polytree-structures. We propose ...
متن کاملImprecise Probabilistic Graphical Models: Equivalent Representations, Inference Algorithms and Applications
Credal networks are probabilistic graphical models that extend Bayesian nets to deal with imprecision in probability, and can actually be regarded as sets of Bayesian nets. Credal nets appear to be powerful means to represent and deal with many important and challenging problems in uncertain reasoning. The counterpart of having more freedom in the modeling phase is an increased inferential comp...
متن کاملEGL2U: Tractable Inference on Large Scale Credal Networks
Credal networks [1, 2] generalize Bayesian networks [3] by associating with variables (closed convex) sets of conditional probability mass functions, i.e., credal sets1, in place of precise conditional probability distributions. Credal networks are models of imprecise probabilities [4], which allow the capturing of incompleteness and imprecision of human knowledge and beliefs [1]. Credal networ...
متن کامل