A Differential Approach to Causality in Staged Trees

Christiane Görgen, Jim Q. Smith
Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:207-215, 2016.

Abstract

In this paper, we apply a recently developed differential approach to inference in staged tree models to causal inference. Staged trees generalise modelling techniques established for Bayesian networks (BN). They have the advantage that they can depict highly nuanced structure impossible to express in a BN and also enable us to perform causal manipulations associated with very general types of interventions on the system. Conveniently, what we call the interpolating polynomial of a staged tree has been found to be an analogue to the essential graph of a BN. By analysing this polynomial in a differential framework, we find that interventions on the model can be expressed as a very simple operation. We can therefore clearly state causal hypotheses which are invariant for all staged trees representing the same causal model. The technology we develop here, illustrated through a simple example, enables us to search for a variety of complex manipulations in large systems accurately and efficiently.

Cite this Paper

BibTeX
@InProceedings{pmlr-v52-goergen16, title = {A Differential Approach to Causality in Staged Trees}, author = {Görgen, Christiane and Smith, Jim Q.}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {207--215}, year = {2016}, editor = {Antonucci, Alessandro and Corani, Giorgio and Campos}, Cassio Polpo}, volume = {52}, series = {Proceedings of Machine Learning Research}, address = {Lugano, Switzerland}, month = {06--09 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v52/goergen16.pdf}, url = {https://proceedings.mlr.press/v52/goergen16.html}, abstract = {In this paper, we apply a recently developed differential approach to inference in staged tree models to causal inference. Staged trees generalise modelling techniques established for Bayesian networks (BN). They have the advantage that they can depict highly nuanced structure impossible to express in a BN and also enable us to perform causal manipulations associated with very general types of interventions on the system. Conveniently, what we call the interpolating polynomial of a staged tree has been found to be an analogue to the essential graph of a BN. By analysing this polynomial in a differential framework, we find that interventions on the model can be expressed as a very simple operation. We can therefore clearly state causal hypotheses which are invariant for all staged trees representing the same causal model. The technology we develop here, illustrated through a simple example, enables us to search for a variety of complex manipulations in large systems accurately and efficiently.} }
Endnote
%0 Conference Paper %T A Differential Approach to Causality in Staged Trees %A Christiane Görgen %A Jim Q. Smith %B Proceedings of the Eighth International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2016 %E Alessandro Antonucci %E Giorgio Corani %E Cassio Polpo Campos} %F pmlr-v52-goergen16 %I PMLR %P 207--215 %U https://proceedings.mlr.press/v52/goergen16.html %V 52 %X In this paper, we apply a recently developed differential approach to inference in staged tree models to causal inference. Staged trees generalise modelling techniques established for Bayesian networks (BN). They have the advantage that they can depict highly nuanced structure impossible to express in a BN and also enable us to perform causal manipulations associated with very general types of interventions on the system. Conveniently, what we call the interpolating polynomial of a staged tree has been found to be an analogue to the essential graph of a BN. By analysing this polynomial in a differential framework, we find that interventions on the model can be expressed as a very simple operation. We can therefore clearly state causal hypotheses which are invariant for all staged trees representing the same causal model. The technology we develop here, illustrated through a simple example, enables us to search for a variety of complex manipulations in large systems accurately and efficiently.
RIS
TY - CPAPER TI - A Differential Approach to Causality in Staged Trees AU - Christiane Görgen AU - Jim Q. Smith BT - Proceedings of the Eighth International Conference on Probabilistic Graphical Models DA - 2016/08/15 ED - Alessandro Antonucci ED - Giorgio Corani ED - Cassio Polpo Campos} ID - pmlr-v52-goergen16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 52 SP - 207 EP - 215 L1 - http://proceedings.mlr.press/v52/goergen16.pdf UR - https://proceedings.mlr.press/v52/goergen16.html AB - In this paper, we apply a recently developed differential approach to inference in staged tree models to causal inference. Staged trees generalise modelling techniques established for Bayesian networks (BN). They have the advantage that they can depict highly nuanced structure impossible to express in a BN and also enable us to perform causal manipulations associated with very general types of interventions on the system. Conveniently, what we call the interpolating polynomial of a staged tree has been found to be an analogue to the essential graph of a BN. By analysing this polynomial in a differential framework, we find that interventions on the model can be expressed as a very simple operation. We can therefore clearly state causal hypotheses which are invariant for all staged trees representing the same causal model. The technology we develop here, illustrated through a simple example, enables us to search for a variety of complex manipulations in large systems accurately and efficiently. ER -
APA
Görgen, C. & Smith, J.Q.. (2016). A Differential Approach to Causality in Staged Trees. Proceedings of the Eighth International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 52:207-215 Available from https://proceedings.mlr.press/v52/goergen16.html.

Related Material