arXiv cs.AI· Frederik Hytting J{\o}rgensen, Luigi Gresele, Sebastian Weichwald·· 12 小时前AI 评分31
因果模型与因果表征究竟“因果”在何处？
What is causal about causal models and representations?
AI 导读
该研究提出一个形式化框架，将因果模型与它试图描述的数据生成过程分离，并形式化“动作到干预”的解释映射。框架内证明了一个不可能性结果：没有任何解释能同时满足一组理想性质且保持非循环；在循环解释下，任何拟合观测分布的模型都具备干预有效性，从而无法被实验证伪。作者分析了若干非循环替代方案，并讨论其对证伪、因果发现、因果表征学习与因果抽象的启示。
正文 · 原文
Abstract:Causal Bayesian networks are often described as causal because they support reasoning about interventions. Yet interventions are defined in the mathematical model, whereas the resulting model predictions concern the effects of actions in the real world. Consequently, whether and how a causal model makes falsifiable predictions about the effects of actions depends on how actions are interpreted as interventions. This requires specifying which actions correspond to which interventions, a step that is frequently left implicit. We develop a formal framework that separates a causal model from the data-generating process it aims to describe and formalizes interpretations connecting actions to interventions. Within this framework, we prove an impossibility result: no interpretation can simultaneously satisfy a collection of desirable properties while remaining non-circular. Under a circular interpretation, every model that fits the observational distribution is interventionally valid, so no experiment can falsify it. We analyze several non-circular alternatives and discuss implications for falsification, causal discovery, causal representation learning, and causal abstraction. More broadly, our framework provides a way to make assumptions about the connections between actions and interventions explicit and to reason formally about the empirical content of causal models.
| Subjects: | Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Statistics Theory (math.ST) |
| Cite as: | arXiv:2501.19335 [stat.ML] |
| (or arXiv:2501.19335v3 [stat.ML] for this version) | |
| https://doi.org/10.48550/arXiv.2501.19335 arXiv-issued DOI via DataCite |
Submission history
From: Frederik Hytting Jørgensen [view email]
[v1]
Fri, 31 Jan 2025 17:35:21 UTC (155 KB)
[v2]
Mon, 3 Feb 2025 17:24:50 UTC (155 KB)
[v3]
Fri, 16 Oct 2026 05:02:31 UTC (160 KB)
来源：arXiv cs.AI · arxiv.org