“SUM-PRODUCT NETWORKS FOR PROBABILISTIC MODELING”

Using a deep network structure, they are able to represent
highly complex variable dependencies, while at the same time many
inference scenarios can be solved with computational costs linear in
the representation size of the SPN.
In this talk, an overview of SPNs is given, following their evolution
in literature. First we define SPNs for discrete random variables and
introduce the notions of completeness, consistency, decomposability and
network polynomials.