Sequential Underspecified Instrument Selection for Cause-Effect Estimation

被引:0
|
作者
Ailer, Elisabeth [1 ,2 ]
Hartford, Jason [3 ,4 ]
Kilbertus, Niki [1 ,2 ,5 ]
机构
[1] Helmholtz Munich, HelmholtzAI, Munich, Germany
[2] Tech Univ Munich, Sch Comp Informat & Technol, Munich, Germany
[3] MILA Quebec Artificial Intelligence Inst, Montreal, PQ, Canada
[4] Recursion, Montreal, PQ, Canada
[5] Munich Ctr Machine Learning, Munich, Germany
基金
加拿大自然科学与工程研究理事会;
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Instrumental variable (IV) methods are used to estimate causal effects in settings with unobserved confounding, where we cannot directly experiment on the treatment variable. Instruments are variables which only affect the outcome indirectly via the treatment variable(s). Most IV applications focus on low-dimensional treatments and crucially require at least as many instruments as treatments. This assumption is restrictive: in the natural sciences we often seek to infer causal effects of high-dimensional treatments (e.g., the effect of gene expressions or microbiota on health and disease), but can only run few experiments with a limited number of instruments (e.g., drugs or antibiotics). In such underspecified problems, the full treatment effect is not identifiable in a single experiment even in the linear case. We show that one can still reliably recover the projection of the treatment effect onto the instrumented subspace and develop techniques to consistently combine such partial estimates from different sets of instruments. We then leverage our combined estimators in an algorithm that iteratively proposes the most informative instruments at each round of experimentation to maximize the overall information about the full causal effect.
引用
收藏
页码:408 / 420
页数:13
相关论文
共 50 条
  • [21] Identifying Crucial Attributes of Sustainable Operation for Scientific Instrument Platforms With a Fuzzy Synthetic Cause-Effect Model
    Wang, Xiaobo
    An, Zhishan
    Zhang, Caixia
    SAGE OPEN, 2023, 13 (03):
  • [22] Towards a Learning Theory of Cause-Effect Inference
    Lopez-Paz, David
    Muandet, Krikamol
    Scholkopf, Bernhard
    Tolstikhin, Ilya
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 37, 2015, 37 : 1452 - 1461
  • [23] Assessment of cause-effect relationships for carcinogenic agents
    Vineis, Paolo
    EPIDEMIOLOGY, 2006, 17 (06) : S47 - S47
  • [24] Environmental epidemiology - Basics and proof of cause-effect
    Weed, DL
    TOXICOLOGY, 2002, 181 : 399 - 403
  • [25] Metabolic Syndrome and Colorectal Neoplasm Cause-Effect or Effect-Cause Relationship
    Li, Tsai-Chung
    Lin, Cheng-Chieh
    JOURNAL OF CLINICAL GASTROENTEROLOGY, 2013, 47 (01) : 6 - 8
  • [26] Environmental epidemiology: basics and proof of cause-effect
    Weed, DL
    TOXICOLOGY, 2001, 164 (1-3) : 29 - 29
  • [27] Cause-Effect Modelling of Soil Liming in Poland
    Zadrozny, Pawel
    Nicia, Pawel
    Wojewodzic, Tomasz
    Dacko, Mariusz
    Paluch, Lukasz
    Plonka, Aleksandra
    Janus, Jaroslaw
    Pijanowski, Jacek
    Bejger, Romualda
    Ukalska-Jaruga, Aleksandra
    Smreczak, Bozena
    Parzych, Piotr
    SUSTAINABILITY, 2024, 16 (23)
  • [28] Cause-effect analysis in assessment of mineral resources
    Sirotinskaya S.V.
    Natural Resources Research, 2004, 13 (1) : 17 - 28
  • [29] Neuropsychological model of thinking in "cause-effect" categories
    Grzegorczyk, Grzegorz
    PSYCHIATRIA I PSYCHOLOGIA KLINICZNA-JOURNAL OF PSYCHIATRY AND CLINICAL PSYCHOLOGY, 2009, 9 (02): : 126 - 130
  • [30] Forest monitoring: Substantiating cause-effect relationships
    Seidling, Walter
    SCIENCE OF THE TOTAL ENVIRONMENT, 2019, 687 : 610 - 617