Learning Causal Structure from Overlapping Variable Sets

Sofia Triantafillou, Ioannis Tsamardinos, Ioannis Tollis
Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, PMLR 9:860-867, 2010.

Abstract

We present an algorithm name cSAT+ for learning the causal structure in a domain from datasets measuring different variables sets. The algorithm outputs a graph with edges corresponding to all possible pairwise causal relations between two variables, named Pairwise Causal Graph (PCG). Examples of interesting inferences include the induction of the absence or presence of some causal relation between two variables never measured together. cSAT+ converts the problem to a series of SAT problems, obtaining leverage from the efficiency of state-of-the-art solvers. In our empirical evaluation, it is shown to outperform ION, the first algorithm solving a similar but more general problem, by two orders of magnitude.

Cite this Paper

BibTeX
@InProceedings{pmlr-v9-triantafillou10a, title = {Learning Causal Structure from Overlapping Variable Sets}, author = {Triantafillou, Sofia and Tsamardinos, Ioannis and Tollis, Ioannis}, booktitle = {Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics}, pages = {860--867}, year = {2010}, editor = {Teh, Yee Whye and Titterington, Mike}, volume = {9}, series = {Proceedings of Machine Learning Research}, address = {Chia Laguna Resort, Sardinia, Italy}, month = {13--15 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v9/triantafillou10a/triantafillou10a.pdf}, url = {https://proceedings.mlr.press/v9/triantafillou10a.html}, abstract = {We present an algorithm name cSAT+ for learning the causal structure in a domain from datasets measuring different variables sets. The algorithm outputs a graph with edges corresponding to all possible pairwise causal relations between two variables, named Pairwise Causal Graph (PCG). Examples of interesting inferences include the induction of the absence or presence of some causal relation between two variables never measured together. cSAT+ converts the problem to a series of SAT problems, obtaining leverage from the efficiency of state-of-the-art solvers. In our empirical evaluation, it is shown to outperform ION, the first algorithm solving a similar but more general problem, by two orders of magnitude.} }
Endnote
%0 Conference Paper %T Learning Causal Structure from Overlapping Variable Sets %A Sofia Triantafillou %A Ioannis Tsamardinos %A Ioannis Tollis %B Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2010 %E Yee Whye Teh %E Mike Titterington %F pmlr-v9-triantafillou10a %I PMLR %P 860--867 %U https://proceedings.mlr.press/v9/triantafillou10a.html %V 9 %X We present an algorithm name cSAT+ for learning the causal structure in a domain from datasets measuring different variables sets. The algorithm outputs a graph with edges corresponding to all possible pairwise causal relations between two variables, named Pairwise Causal Graph (PCG). Examples of interesting inferences include the induction of the absence or presence of some causal relation between two variables never measured together. cSAT+ converts the problem to a series of SAT problems, obtaining leverage from the efficiency of state-of-the-art solvers. In our empirical evaluation, it is shown to outperform ION, the first algorithm solving a similar but more general problem, by two orders of magnitude.
RIS
TY - CPAPER TI - Learning Causal Structure from Overlapping Variable Sets AU - Sofia Triantafillou AU - Ioannis Tsamardinos AU - Ioannis Tollis BT - Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics DA - 2010/03/31 ED - Yee Whye Teh ED - Mike Titterington ID - pmlr-v9-triantafillou10a PB - PMLR DP - Proceedings of Machine Learning Research VL - 9 SP - 860 EP - 867 L1 - http://proceedings.mlr.press/v9/triantafillou10a/triantafillou10a.pdf UR - https://proceedings.mlr.press/v9/triantafillou10a.html AB - We present an algorithm name cSAT+ for learning the causal structure in a domain from datasets measuring different variables sets. The algorithm outputs a graph with edges corresponding to all possible pairwise causal relations between two variables, named Pairwise Causal Graph (PCG). Examples of interesting inferences include the induction of the absence or presence of some causal relation between two variables never measured together. cSAT+ converts the problem to a series of SAT problems, obtaining leverage from the efficiency of state-of-the-art solvers. In our empirical evaluation, it is shown to outperform ION, the first algorithm solving a similar but more general problem, by two orders of magnitude. ER -
APA
Triantafillou, S., Tsamardinos, I. & Tollis, I.. (2010). Learning Causal Structure from Overlapping Variable Sets. Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 9:860-867 Available from https://proceedings.mlr.press/v9/triantafillou10a.html.

Related Material