Bayesian network structure learning with causal effects in the presence of latent variables

Kiattikun Chobtham, Anthony C. Constantinou
Proceedings of the 10th International Conference on Probabilistic Graphical Models, PMLR 138:101-112, 2020.

Abstract

Latent variables may lead to spurious relationships that can be misinterpreted as causal relationships. In Bayesian Networks (BNs), this challenge is known as learning under causal insufficiency. Structure learning algorithms that assume causal insufficiency tend to reconstruct the ancestral graph of a BN, where bi-directed edges represent confounding and directed edges represent direct or ancestral relationships. This paper describes a hybrid structure learning algorithm, called CCHM, which combines the constraint-based part of cFCI with hill-climbing score-based learning. The score-based process incorporates Pearl’s do-calculus to measure causal effects, which are used to orientate edges that would otherwise remain undirected, under the assumption the BN is a linear Structure Equation Model where data follow a multivariate Gaussian distribution. Experiments based on both randomised and well-known networks show that CCHM improves the state-of-the-art in terms of reconstructing the true ancestral graph.

Cite this Paper

BibTeX
@InProceedings{pmlr-v138-chobtham20a, title = { Bayesian network structure learning with causal effects in the presence of latent variables }, author = {Chobtham, Kiattikun and Constantinou, Anthony C.}, booktitle = {Proceedings of the 10th International Conference on Probabilistic Graphical Models}, pages = {101--112}, year = {2020}, editor = {Jaeger, Manfred and Nielsen, Thomas Dyhre}, volume = {138}, series = {Proceedings of Machine Learning Research}, month = {23--25 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v138/chobtham20a/chobtham20a.pdf}, url = {https://proceedings.mlr.press/v138/chobtham20a.html}, abstract = { Latent variables may lead to spurious relationships that can be misinterpreted as causal relationships. In Bayesian Networks (BNs), this challenge is known as learning under causal insufficiency. Structure learning algorithms that assume causal insufficiency tend to reconstruct the ancestral graph of a BN, where bi-directed edges represent confounding and directed edges represent direct or ancestral relationships. This paper describes a hybrid structure learning algorithm, called CCHM, which combines the constraint-based part of cFCI with hill-climbing score-based learning. The score-based process incorporates Pearl’s do-calculus to measure causal effects, which are used to orientate edges that would otherwise remain undirected, under the assumption the BN is a linear Structure Equation Model where data follow a multivariate Gaussian distribution. Experiments based on both randomised and well-known networks show that CCHM improves the state-of-the-art in terms of reconstructing the true ancestral graph.} }
Endnote
%0 Conference Paper %T Bayesian network structure learning with causal effects in the presence of latent variables %A Kiattikun Chobtham %A Anthony C. Constantinou %B Proceedings of the 10th International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2020 %E Manfred Jaeger %E Thomas Dyhre Nielsen %F pmlr-v138-chobtham20a %I PMLR %P 101--112 %U https://proceedings.mlr.press/v138/chobtham20a.html %V 138 %X Latent variables may lead to spurious relationships that can be misinterpreted as causal relationships. In Bayesian Networks (BNs), this challenge is known as learning under causal insufficiency. Structure learning algorithms that assume causal insufficiency tend to reconstruct the ancestral graph of a BN, where bi-directed edges represent confounding and directed edges represent direct or ancestral relationships. This paper describes a hybrid structure learning algorithm, called CCHM, which combines the constraint-based part of cFCI with hill-climbing score-based learning. The score-based process incorporates Pearl’s do-calculus to measure causal effects, which are used to orientate edges that would otherwise remain undirected, under the assumption the BN is a linear Structure Equation Model where data follow a multivariate Gaussian distribution. Experiments based on both randomised and well-known networks show that CCHM improves the state-of-the-art in terms of reconstructing the true ancestral graph.
APA
Chobtham, K. & Constantinou, A.C.. (2020). Bayesian network structure learning with causal effects in the presence of latent variables . Proceedings of the 10th International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 138:101-112 Available from https://proceedings.mlr.press/v138/chobtham20a.html.

Related Material