论文概要
研究领域: ML 作者: Rafal Urbaniak, Sam Witty, Daniel Waxman 发布时间: 2026-09-06 arXiv: 2509.04285
中文摘要
解释特定结果为何发生、哪些输入应被归责或归功于,是哲学、科学和政策分析的核心问题。现有工具分为两派:实际因果理论(AC)提供原则性判断,但仅适用于玩具级模型,因为其计算需要枚举反事实场景;可扩展归因方法如SHAP(甚至因果SHAP)至少部分忽略了生成数据的因果结构,可能给出与审慎因果分析冲突的答案。我们通过概率因果影响(PCI)框架弥合了这一鸿沟。PCI建立在实际因果性和Pearl的必要性与充分性概率概念之上,但将可解释性问题重构为概率因果模型上的估计问题,可通过蒙特卡洛轻松近似。通过指定候选解释分布、反事实值分布和评分函数,PCI提供了可处理的、有因果基础的、分级的解释,将AC和Pearl的因果概率作为退化情形加以推广。我们在合成和真实世界示例中评估了PCI,包括与AC的一致性检验、扩展实验、复杂连续值动态系统,以及一个在数百万数据点上训练的因果机器学习模型。
原文摘要
Explaining why a specific outcome occurred, and which inputs deserve the blame or credit, is central to philosophical, scientific, and policy analysis. Existing tools split into two camps. The theory of actual causality (AC) gives principled verdicts, but only for toy-sized models, because computing them requires enumerating counterfactual scenarios. Scalable attribution methods like SHAP (or even causal SHAP) at least partially ignore the causal structure that generated the data, and can give answers that conflict with a careful causal analysis. We close this gap with Probabilistic Causal Impact (PCI). PCI builds on actual causality and on Pearl’s notions of probability of necessity and sufficiency, but recasts the question of explainability as an estimation problem on a probabilistic cau…
— 自动采集于 2026-09-07
#论文 #arXiv #ML #小凯
