Adjustment Identification Distance: A gadjid for Causal Structure Learning

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

2 Downloads (Pure)

Abstract

Evaluating graphs learned by causal discovery algorithms is difficult: The number of edges that differ between two graphs does not reflect how the graphs differ with respect to the identifying formulas they suggest for causal effects. We introduce a framework for developing causal distances between graphs which includes the structural intervention distance for directed acyclic graphs as a special case. We use this framework to develop improved adjustment-based distances as well as extensions to completed partially directed acyclic graphs and causal orders. We develop new reachability algorithms to compute the distances efficiently and to prove their low polynomial time complexity. In our package gadjid (open source at github.com/CausalDisco/gadjid), we provide implementations of our distances; they are orders of magnitude faster with proven lower time complexity than the structural intervention distance and thereby provide a success metric for causal discovery that scales to graph sizes that were previously prohibitive.

Original languageEnglish
Title of host publicationProceedings of the 40th Conference on Uncertainty in Artificial Intelligence (UAI 2024)
Number of pages30
Volume244
PublisherPMLR
Publication date2024
Pages1569-1598
Publication statusPublished - 2024
Event40th Conference on Uncertainty in Artificial Intelligence, UAI 2024 - Barcelona, Spain
Duration: 15 Jul 202419 Jul 2024

Conference

Conference40th Conference on Uncertainty in Artificial Intelligence, UAI 2024
Country/TerritorySpain
CityBarcelona
Period15/07/202419/07/2024
SponsorBarcelona School of Economics (BSE), DE Shaw and Co, Google, Huawei, Universitat Pompeu Fabra (UPF)
SeriesProceedings of Machine Learning Research
Volume244
ISSN2640-3498

Bibliographical note

Publisher Copyright:
© 2024 Proceedings of Machine Learning Research. All rights reserved.

Cite this