Causal Discovery from Subsampled Time Series Data by Constraint Optimization

Antti Hyttinen, Sergey Plis, Matti Järvisalo, Frederick Eberhardt, David Danks
Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:216-227, 2016.

Abstract

This paper focuses on causal structure estimation from time series data in which measurements are obtained at a coarser timescale than the causal timescale of the underlying system. Previous work has shown that such subsampling can lead to significant errors about the system’s causal structure if not properly taken into account. In this paper, we first consider the search for the system timescale causal structures that correspond to a given measurement timescale structure. We provide a constraint satisfaction procedure whose computational performance is several orders of magnitude better than previous approaches. We then consider finite-sample data as input, and propose the first constraint optimization approach for recovering the system timescale causal structure. This algorithm optimally recovers from possible conflicts due to statistical errors. More generally, these advances allow for a robust and non-parametric estimation of system timescale causal structures from subsampled time series data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v52-hyttinen16, title = {Causal Discovery from Subsampled Time Series Data by Constraint Optimization}, author = {Hyttinen, Antti and Plis, Sergey and Järvisalo, Matti and Eberhardt, Frederick and Danks, David}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {216--227}, 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/hyttinen16.pdf}, url = {https://proceedings.mlr.press/v52/hyttinen16.html}, abstract = {This paper focuses on causal structure estimation from time series data in which measurements are obtained at a coarser timescale than the causal timescale of the underlying system. Previous work has shown that such subsampling can lead to significant errors about the system’s causal structure if not properly taken into account. In this paper, we first consider the search for the system timescale causal structures that correspond to a given measurement timescale structure. We provide a constraint satisfaction procedure whose computational performance is several orders of magnitude better than previous approaches. We then consider finite-sample data as input, and propose the first constraint optimization approach for recovering the system timescale causal structure. This algorithm optimally recovers from possible conflicts due to statistical errors. More generally, these advances allow for a robust and non-parametric estimation of system timescale causal structures from subsampled time series data.} }
Endnote
%0 Conference Paper %T Causal Discovery from Subsampled Time Series Data by Constraint Optimization %A Antti Hyttinen %A Sergey Plis %A Matti Järvisalo %A Frederick Eberhardt %A David Danks %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-hyttinen16 %I PMLR %P 216--227 %U https://proceedings.mlr.press/v52/hyttinen16.html %V 52 %X This paper focuses on causal structure estimation from time series data in which measurements are obtained at a coarser timescale than the causal timescale of the underlying system. Previous work has shown that such subsampling can lead to significant errors about the system’s causal structure if not properly taken into account. In this paper, we first consider the search for the system timescale causal structures that correspond to a given measurement timescale structure. We provide a constraint satisfaction procedure whose computational performance is several orders of magnitude better than previous approaches. We then consider finite-sample data as input, and propose the first constraint optimization approach for recovering the system timescale causal structure. This algorithm optimally recovers from possible conflicts due to statistical errors. More generally, these advances allow for a robust and non-parametric estimation of system timescale causal structures from subsampled time series data.
RIS
TY - CPAPER TI - Causal Discovery from Subsampled Time Series Data by Constraint Optimization AU - Antti Hyttinen AU - Sergey Plis AU - Matti Järvisalo AU - Frederick Eberhardt AU - David Danks 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-hyttinen16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 52 SP - 216 EP - 227 L1 - http://proceedings.mlr.press/v52/hyttinen16.pdf UR - https://proceedings.mlr.press/v52/hyttinen16.html AB - This paper focuses on causal structure estimation from time series data in which measurements are obtained at a coarser timescale than the causal timescale of the underlying system. Previous work has shown that such subsampling can lead to significant errors about the system’s causal structure if not properly taken into account. In this paper, we first consider the search for the system timescale causal structures that correspond to a given measurement timescale structure. We provide a constraint satisfaction procedure whose computational performance is several orders of magnitude better than previous approaches. We then consider finite-sample data as input, and propose the first constraint optimization approach for recovering the system timescale causal structure. This algorithm optimally recovers from possible conflicts due to statistical errors. More generally, these advances allow for a robust and non-parametric estimation of system timescale causal structures from subsampled time series data. ER -
APA
Hyttinen, A., Plis, S., Järvisalo, M., Eberhardt, F. & Danks, D.. (2016). Causal Discovery from Subsampled Time Series Data by Constraint Optimization. Proceedings of the Eighth International Conference on Probabilistic Graphical Models, in Proceedings of Machine Learning Research 52:216-227 Available from https://proceedings.mlr.press/v52/hyttinen16.html.

Related Material