Process Independence Testing in Proximal Graphical Event Models

Debarun Bhattacharjya, Karthikeyan Shanmugam, Tian Gao, Dharmashankar Subramanian
Proceedings of the First Conference on Causal Learning and Reasoning, PMLR 177:144-161, 2022.

Abstract

Datasets involving irregular occurrences of different types of events over the timeline are increasingly commonly available. Proximal graphical event models (PGEMs) are a recent graphical representation for modeling relationships between different event types in such datasets. Existing algorithms for learning PGEMs from event datasets perform poorly on the task of structure discovery, which is particularly important for causal inference since the underlying graph determines the effect of interventions. In this paper, we explore causal semantics in PGEMs and study process independencies implied by the graphical structure of the model. We introduce (conditional) process independence tests for causal PGEMs, deploying them using variations of constraint-based structure discovery algorithms for Bayesian networks. Through experiments with synthetic and real datasets, we show that the proposed approaches are better at balancing precision and recall, demonstrating improved F1 scores over state-of-the-art baselines.

Cite this Paper

BibTeX
@InProceedings{pmlr-v177-bhattacharjya22a, title = {Process Independence Testing in Proximal Graphical Event Models}, author = {Bhattacharjya, Debarun and Shanmugam, Karthikeyan and Gao, Tian and Subramanian, Dharmashankar}, booktitle = {Proceedings of the First Conference on Causal Learning and Reasoning}, pages = {144--161}, year = {2022}, editor = {Schölkopf, Bernhard and Uhler, Caroline and Zhang, Kun}, volume = {177}, series = {Proceedings of Machine Learning Research}, month = {11--13 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v177/bhattacharjya22a/bhattacharjya22a.pdf}, url = {https://proceedings.mlr.press/v177/bhattacharjya22a.html}, abstract = {Datasets involving irregular occurrences of different types of events over the timeline are increasingly commonly available. Proximal graphical event models (PGEMs) are a recent graphical representation for modeling relationships between different event types in such datasets. Existing algorithms for learning PGEMs from event datasets perform poorly on the task of structure discovery, which is particularly important for causal inference since the underlying graph determines the effect of interventions. In this paper, we explore causal semantics in PGEMs and study process independencies implied by the graphical structure of the model. We introduce (conditional) process independence tests for causal PGEMs, deploying them using variations of constraint-based structure discovery algorithms for Bayesian networks. Through experiments with synthetic and real datasets, we show that the proposed approaches are better at balancing precision and recall, demonstrating improved F1 scores over state-of-the-art baselines.} }
Endnote
%0 Conference Paper %T Process Independence Testing in Proximal Graphical Event Models %A Debarun Bhattacharjya %A Karthikeyan Shanmugam %A Tian Gao %A Dharmashankar Subramanian %B Proceedings of the First Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2022 %E Bernhard Schölkopf %E Caroline Uhler %E Kun Zhang %F pmlr-v177-bhattacharjya22a %I PMLR %P 144--161 %U https://proceedings.mlr.press/v177/bhattacharjya22a.html %V 177 %X Datasets involving irregular occurrences of different types of events over the timeline are increasingly commonly available. Proximal graphical event models (PGEMs) are a recent graphical representation for modeling relationships between different event types in such datasets. Existing algorithms for learning PGEMs from event datasets perform poorly on the task of structure discovery, which is particularly important for causal inference since the underlying graph determines the effect of interventions. In this paper, we explore causal semantics in PGEMs and study process independencies implied by the graphical structure of the model. We introduce (conditional) process independence tests for causal PGEMs, deploying them using variations of constraint-based structure discovery algorithms for Bayesian networks. Through experiments with synthetic and real datasets, we show that the proposed approaches are better at balancing precision and recall, demonstrating improved F1 scores over state-of-the-art baselines.
APA
Bhattacharjya, D., Shanmugam, K., Gao, T. & Subramanian, D.. (2022). Process Independence Testing in Proximal Graphical Event Models. Proceedings of the First Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 177:144-161 Available from https://proceedings.mlr.press/v177/bhattacharjya22a.html.

Related Material