arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2009.09768cs.AIcs.LG
收藏

通过上下文特定独立关系识别因果效应

Identifying Causal Effects via Context-specific Independence Relations

University of Jyvaskyla(于韦斯屈莱大学);University of Helsinki(赫尔辛基大学)

Santtu Tikka, Antti Hyttinen, Juha Karvanen

AI总结:

本文针对存在上下文特定独立关系时因果效应识别不完备的问题,提出一种可证明可靠且包含标准 do-calculus 的演算与自动搜索方法,并证明少量 CSI 关系即可使原本不可识别的实例变得可识别。

摘要:

因果效应识别考虑的是,在给定的因果结构中,干预概率分布能否由被动观测到的分布唯一确定。如果生成系统诱导出上下文特定独立(CSI)关系,那么现有的基于 do-calculus 的识别程序和准则本质上是不完备的。我们证明,在存在 CSI 的情况下,判定因果效应不可识别性是 NP-hard 的。受此启发,我们设计了一种用于在存在 CSI 的情况下识别因果效应的演算和自动搜索程序。该方法在理论上被证明是可靠的,并且将标准 do-calculus 作为一个特例包含在内。借助该方法,我们能够获得以前无法得到的识别公式,并证明少量的 CSI 关系可能足以将一个以前不可识别的实例转变为可识别的。

英文摘要:

Causal effect identification considers whether an interventional probability distribution can be uniquely determined from a passively observed distribution in a given causal structure. If the generating system induces context-specific independence (CSI) relations, the existing identification procedures and criteria based on do-calculus are inherently incomplete. We show that deciding causal effect non-identifiability is NP-hard in the presence of CSIs. Motivated by this, we design a calculus and an automated search procedure for identifying causal effects in the presence of CSIs. The approach is provably sound and it includes standard do-calculus as a special case. With the approach we can obtain identifying formulas that were unobtainable previously, and demonstrate that a small number of CSI-relations may be sufficient to turn a previously non-identifiable instance to identifiable.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑