arXiv：cs.LG· Uzair Akbar, Zulfiqar Zaidi, Niki Kilbertus, Krikamol Muandet, Bo Dai·· 4 小时前AI 评分33
Symmetry-Informed Causal Partial Identification：利用数据对称性收紧因果部分识别边界
Symmetry-Informed Causal Partial Identification
AI 导读
研究者提出将已知数据对称性——因果效应在某些数据变换下的不变性——作为部分识别（PI）的新约束来源，通过将其转化为因果函数的形状约束并配合测度变换，在两种经典 PI 模型下收紧因果效应边界。该结果在总体情形得到理论证明，并在有限样本实验中验证。
正文
Abstract:Partial identification (PI) entails estimating bounds on causal effects by encoding different assumptions on data generation as a constrained optimization problem. Such bounds can suffice to inform policy decisions even if the causal effect itself is not identifiable. Often vacuous in practice, practitioners seek to exhaustively encode domain knowledge as additional constraints to make the PI bounds more informative. We introduce known data symmetries -- invariance of the causal effect under certain data transformations -- as a new source of constraints to inform PI. We operationalize this as a shape constraint on the causal function, and via a change of measure against which PI is posed using simple data pre-processing. Both approaches are shown to sharpen bounds under two canonical PI models. This is shown both theoretically for the population case, and via experiments in the finite-sample case. More broadly, our framework establishes data symmetries as a natural, underutilized source of background knowledge for robust causal inference.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.09230 [cs.LG] |
| (or arXiv:2610.09230v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09230 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zulfiqar Zaidi [view email]
[v1]
Tue, 6 Oct 2026 23:46:49 UTC (1,059 KB)
来源：arXiv：cs.LG · arxiv.org