A minimal approach to causal inference on topologies with bounded indegree

The structure of the causal interdependencies between processes in a causal, stochastic dynamical system can be succinctly characterized by a generative model. Inferring the structure of the generative model, however, requires calculating divergences using the full joint statistics. For the case when an upperbound on the indegree of each process is known, we describe a computationally efficient method using directed information which does not require the full statistics and recovers the parents of each process independently from finding the parents of other processes.