GRAPHICAL METHODS FOR INEQUALITY CONSTRAINTS IN MARGINALIZED DAGS

Journal article

We present a graphical approach to deriving inequality constraints for directed acyclic graph (DAG) models, where some variables are unobserved. In particular we show that the observed distribution of a discrete model is always restricted if any two observed variables are neither adjacent in the graph, nor share a latent parent; this generalizes the well known instrumental inequality. The method also provides inequalities on interventional distributions, which can be used to bound causal effects. All these constraints are characterized in terms of a new graphical separation criterion, providing an easy and intuitive method for their derivation. © 2012 IEEE.

Authors: Evans, R

• Publication status: Published
• Journal: 2012 IEEE INTERNATIONAL WORKSHOP ON MACHINE LEARNING FOR SIGNAL PROCESSING (MLSP)
• Publication date: 2012-01-01
• DOI: 10.1109/MLSP.2012.6349796
• EISSN: 2161-0371
• ISSN: 2161-0363
• Language: English
• Keywords: controlled direct effect; intervention; marginalized; directed acyclic graph; Causal model