Sampling graphs from a probabilistic generative model

In this paper we present a method of sampling from a probabilistic generative model for a set of graphs. Our method is based on the assumption that the nodes and edges of graphs arise under independent Bernoulli distributions. We sample graphs from the generative model according to the node and edge occurrence probabilities. We explain the construction of our generative model and then compute the node and edge occurrence probabilities which allow us to formulate a sampling procedure. We demonstrate experimentally to what extent the graphs sampled by our method reproduce the salient properties of the graphs in the original training sample.