Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination

Kun ZHANG, Biwei HUANG, Jiji ZHANG, Clark GLYMOUR, Bernhard SCHÖLKOPF

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)Researchpeer-review

7 Citations (Scopus)

Abstract

It is commonplace to encounter nonstationary or heterogeneous data, of which the underlying generating process changes over time or across data sets (the data sets may have different experimental conditions or data collection conditions). Such a distribution shift feature presents both challenges and opportunities for causal discovery. In this paper we develop a principled framework for causal discovery from such data, called Constraint-based causal Discovery from Nonstationary/heterogeneous Data (CD-NOD), which addresses two important questions. First, we propose an enhanced constraint-based procedure to detect variables whose local mechanisms change and recover the skeleton of the causal structure over observed variables. Second, we present a way to determine causal orientations by making use of independence changes in the data distribution implied by the underlying causal model, benefiting from information carried by changing distributions. Experimental results on various synthetic and real-world data sets are presented to demonstrate the efficacy of our methods.
Original languageEnglish
Title of host publicationProceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence
PublisherInternational Joint Conferences on Artificial Intelligence
Pages1347-1353
Number of pages7
ISBN (Print)9780999241103
DOIs
Publication statusPublished - 1 Jan 2017

Keywords

  • Change and Causality Uncertainty in AI: Uncertainty in AI
  • Knowledge Representation
  • Reasoning
  • and Logic: Action

Cite this

ZHANG, K., HUANG, B., ZHANG, J., GLYMOUR, C., & SCHÖLKOPF, B. (2017). Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (pp. 1347-1353). International Joint Conferences on Artificial Intelligence. https://doi.org/10.24963/ijcai.2017/187
ZHANG, Kun ; HUANG, Biwei ; ZHANG, Jiji ; GLYMOUR, Clark ; SCHÖLKOPF, Bernhard. / Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination. Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence, 2017. pp. 1347-1353
@inproceedings{44ac4c59b22c4c5ca4ac8c16f8577936,
title = "Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination",
abstract = "It is commonplace to encounter nonstationary or heterogeneous data, of which the underlying generating process changes over time or across data sets (the data sets may have different experimental conditions or data collection conditions). Such a distribution shift feature presents both challenges and opportunities for causal discovery. In this paper we develop a principled framework for causal discovery from such data, called Constraint-based causal Discovery from Nonstationary/heterogeneous Data (CD-NOD), which addresses two important questions. First, we propose an enhanced constraint-based procedure to detect variables whose local mechanisms change and recover the skeleton of the causal structure over observed variables. Second, we present a way to determine causal orientations by making use of independence changes in the data distribution implied by the underlying causal model, benefiting from information carried by changing distributions. Experimental results on various synthetic and real-world data sets are presented to demonstrate the efficacy of our methods.",
keywords = "Change and Causality Uncertainty in AI: Uncertainty in AI, Knowledge Representation, Reasoning, and Logic: Action",
author = "Kun ZHANG and Biwei HUANG and Jiji ZHANG and Clark GLYMOUR and Bernhard SCH{\"O}LKOPF",
year = "2017",
month = "1",
day = "1",
doi = "10.24963/ijcai.2017/187",
language = "English",
isbn = "9780999241103",
pages = "1347--1353",
booktitle = "Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence",
publisher = "International Joint Conferences on Artificial Intelligence",

}

ZHANG, K, HUANG, B, ZHANG, J, GLYMOUR, C & SCHÖLKOPF, B 2017, Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination. in Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence, pp. 1347-1353. https://doi.org/10.24963/ijcai.2017/187

Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination. / ZHANG, Kun; HUANG, Biwei; ZHANG, Jiji; GLYMOUR, Clark; SCHÖLKOPF, Bernhard.

Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence, 2017. p. 1347-1353.

Research output: Book Chapters | Papers in Conference ProceedingsConference paper (refereed)Researchpeer-review

TY - GEN

T1 - Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination

AU - ZHANG, Kun

AU - HUANG, Biwei

AU - ZHANG, Jiji

AU - GLYMOUR, Clark

AU - SCHÖLKOPF, Bernhard

PY - 2017/1/1

Y1 - 2017/1/1

N2 - It is commonplace to encounter nonstationary or heterogeneous data, of which the underlying generating process changes over time or across data sets (the data sets may have different experimental conditions or data collection conditions). Such a distribution shift feature presents both challenges and opportunities for causal discovery. In this paper we develop a principled framework for causal discovery from such data, called Constraint-based causal Discovery from Nonstationary/heterogeneous Data (CD-NOD), which addresses two important questions. First, we propose an enhanced constraint-based procedure to detect variables whose local mechanisms change and recover the skeleton of the causal structure over observed variables. Second, we present a way to determine causal orientations by making use of independence changes in the data distribution implied by the underlying causal model, benefiting from information carried by changing distributions. Experimental results on various synthetic and real-world data sets are presented to demonstrate the efficacy of our methods.

AB - It is commonplace to encounter nonstationary or heterogeneous data, of which the underlying generating process changes over time or across data sets (the data sets may have different experimental conditions or data collection conditions). Such a distribution shift feature presents both challenges and opportunities for causal discovery. In this paper we develop a principled framework for causal discovery from such data, called Constraint-based causal Discovery from Nonstationary/heterogeneous Data (CD-NOD), which addresses two important questions. First, we propose an enhanced constraint-based procedure to detect variables whose local mechanisms change and recover the skeleton of the causal structure over observed variables. Second, we present a way to determine causal orientations by making use of independence changes in the data distribution implied by the underlying causal model, benefiting from information carried by changing distributions. Experimental results on various synthetic and real-world data sets are presented to demonstrate the efficacy of our methods.

KW - Change and Causality Uncertainty in AI: Uncertainty in AI

KW - Knowledge Representation

KW - Reasoning

KW - and Logic: Action

UR - http://commons.ln.edu.hk/sw_master/6103

U2 - 10.24963/ijcai.2017/187

DO - 10.24963/ijcai.2017/187

M3 - Conference paper (refereed)

C2 - 28966540

SN - 9780999241103

SP - 1347

EP - 1353

BT - Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence

PB - International Joint Conferences on Artificial Intelligence

ER -

ZHANG K, HUANG B, ZHANG J, GLYMOUR C, SCHÖLKOPF B. Causal discovery from nonstationary/heterogeneous data : skeleton estimation and orientation determination. In Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence. 2017. p. 1347-1353 https://doi.org/10.24963/ijcai.2017/187