On Searching for Generalized Instrumental Variables

Benito Zander; Maciej Liskiewicz

On Searching for Generalized Instrumental Variables

Institute of Theoretical Computer Science

Abstract

Instrumental Variables are a popular way to identify the direct causal effect of a random variable X on a variable Y. Often no single instrumental variable exists, although it is still possible to find a set of generalized instrumental variables (GIVs) and identify the causal effect of all these variables at once. Till now it was not known how to find GIVs systematically or even test efficiently, if given variables satisfy GIV conditions. We provide fast algorithms for searching and testing restricted cases of GIVs. However, we prove that in the most general case it is NP-hard to verify if given variables fulfill the conditions of a general instrumental sets.

Original language	English
Title of host publication	Proceedings of the 19th International Conference on Artificial Intelligence and Statistics
Editors	Arthur Gretton, Christian C. Robert
Number of pages	9
Volume	51
Place of Publication	Cadiz, Spain
Publisher	PMLR
Publication date	01.09.2016
Pages	1214-1222
Publication status	Published - 01.09.2016
Event	Proceedings of the 19th International Conference on Artificial Intelligence and Statistics - Cadiz, Spain Duration: 09.03.2016 → 11.03.2016

Access to Document

Causality: an algorithmic framework and a computational complexity perspective
Liskiewicz, M. & Textor, J.
01.01.16 → 31.12.22
Project: DFG Projects › DFG Individual Projects

Cite this

@inproceedings{d8aaf9775fd9415a8ccf513dbc63ef96,

title = "On Searching for Generalized Instrumental Variables",

abstract = "Instrumental Variables are a popular way to identify the direct causal effect of a random variable X on a variable Y. Often no single instrumental variable exists, although it is still possible to find a set of generalized instrumental variables (GIVs) and identify the causal effect of all these variables at once. Till now it was not known how to find GIVs systematically or even test efficiently, if given variables satisfy GIV conditions. We provide fast algorithms for searching and testing restricted cases of GIVs. However, we prove that in the most general case it is NP-hard to verify if given variables fulfill the conditions of a general instrumental sets.",

author = "Benito Zander and Maciej Liskiewicz",

year = "2016",

month = sep,

day = "1",

language = "English",

volume = "51",

series = "Proceedings of Machine Learning Research",

publisher = "PMLR",

pages = "1214--1222",

editor = "Arthur Gretton and Robert, {Christian C.}",

booktitle = "Proceedings of the 19th International Conference on Artificial Intelligence and Statistics",

note = "Proceedings of the 19th International Conference on Artificial Intelligence and Statistics ; Conference date: 09-03-2016 Through 11-03-2016",

}

Zander, B & Liskiewicz, M 2016, On Searching for Generalized Instrumental Variables. in A Gretton & CC Robert (eds), Proceedings of the 19th International Conference on Artificial Intelligence and Statistics. vol. 51, Proceedings of Machine Learning Research, PMLR, Cadiz, Spain, pp. 1214-1222, Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, Cadiz, Spain, 09.03.16. <http://proceedings.mlr.press/v51/vanderzander16.pdf>

On Searching for Generalized Instrumental Variables. / Zander, Benito ; Liskiewicz, Maciej.
Proceedings of the 19th International Conference on Artificial Intelligence and Statistics. ed. / Arthur Gretton; Christian C. Robert. Vol. 51 Cadiz, Spain: PMLR, 2016. p. 1214-1222 (Proceedings of Machine Learning Research).

Research output: Chapters in Books/Reports/Conference Proceedings › Conference contribution › peer-review

TY - GEN

T1 - On Searching for Generalized Instrumental Variables

AU - Zander, Benito

AU - Liskiewicz, Maciej

PY - 2016/9/1

Y1 - 2016/9/1

N2 - Instrumental Variables are a popular way to identify the direct causal effect of a random variable X on a variable Y. Often no single instrumental variable exists, although it is still possible to find a set of generalized instrumental variables (GIVs) and identify the causal effect of all these variables at once. Till now it was not known how to find GIVs systematically or even test efficiently, if given variables satisfy GIV conditions. We provide fast algorithms for searching and testing restricted cases of GIVs. However, we prove that in the most general case it is NP-hard to verify if given variables fulfill the conditions of a general instrumental sets.

AB - Instrumental Variables are a popular way to identify the direct causal effect of a random variable X on a variable Y. Often no single instrumental variable exists, although it is still possible to find a set of generalized instrumental variables (GIVs) and identify the causal effect of all these variables at once. Till now it was not known how to find GIVs systematically or even test efficiently, if given variables satisfy GIV conditions. We provide fast algorithms for searching and testing restricted cases of GIVs. However, we prove that in the most general case it is NP-hard to verify if given variables fulfill the conditions of a general instrumental sets.

M3 - Conference contribution

VL - 51

T3 - Proceedings of Machine Learning Research

SP - 1214

EP - 1222

BT - Proceedings of the 19th International Conference on Artificial Intelligence and Statistics

A2 - Gretton, Arthur

A2 - Robert, Christian C.

PB - PMLR

CY - Cadiz, Spain

T2 - Proceedings of the 19th International Conference on Artificial Intelligence and Statistics

Y2 - 9 March 2016 through 11 March 2016

ER -

On Searching for Generalized Instrumental Variables

Abstract

Access to Document

Fingerprint

Projects

Causality: an algorithmic framework and a computational complexity perspective

Cite this