Identifying Independencies in Causal Graphs with Feedback

TL;DR

This paper demonstrates that the d-separation criterion can reliably test for conditional independence in feedback systems with discrete variables, expanding causal inference tools.

Contribution

It establishes the validity of d-separation for feedback systems with discrete variables, a previously uncertain area.

Findings

d-separation is valid for feedback systems with discrete variables

Conditional independence can be tested using d-separation in such systems

Supports causal analysis in complex feedback networks

Abstract

We show that the d -separation criterion constitutes a valid test for conditional independence relationships that are induced by feedback systems involving discrete variables.