BriefGPT.xyz
Nov, 2023
潜在类别混淆下的因果发现
Causal Discovery under Latent Class Confounding
HTML
PDF
Bijan Mazaheri, Spencer Gordon, Yuval Rabani, Leonard Schulman
TL;DR
使用有向无环图来建模系统的因果结构。在多个数据源(群体或环境)的数据聚合中,全局混淆模糊了许多因果发现算法中的条件独立性属性。因此,现有的因果发现算法不适用于多源设置。我们证明,如果混淆的基数有限(即数据来自有限数量的源),仍然可以实现因果发现。该问题的可行性取决于全局混淆因素的基数、观测变量的基数和因果结构的稀疏程度的权衡。
Abstract
directed acyclic graphs
are used to model the causal structure of a system. ``
causal discovery
'' describes the problem of learning this structure from data. When data is an aggregate from multiple sources (popula
→