Toward Learning Graphical and Causal Process Models

We describe an approach to learning causal models that leverages temporal information. We posit the existence of a graphical description of a causal process that generates observations through time. We explore assumptions connecting the graphical description with the statistical process and what one can infer about the causal structure of the process under these assumptions.