📄 中文摘要
因果模型的抽象化允许对模型进行粗化，同时保持因果关系。研究提出了一种因果嵌入框架，使得多个详细模型能够映射到一个粗因果模型的子系统中。因果嵌入被定义为抽象的推广，并提出了一种广义的一致性概念。通过定义多分辨率边际问题，展示了因果嵌入在统计边际问题和因果边际问题中的相关性，同时还说明了其在合并来自不同表示模型的数据集中的实际应用。
多层次因果嵌入
出处: Multi-Level Causal Embeddings
发布: 2026年2月27日
📄 English Summary
Multi-Level Causal Embeddings
Causal model abstractions allow for the coarsening of models while preserving cause-and-effect relationships. A framework for causal embeddings is proposed, enabling multiple detailed models to be mapped into sub-systems of a coarser causal model. Causal embeddings are defined as a generalization of abstraction, and a generalized notion of consistency is presented. By defining a multi-resolution marginal problem, the relevance of causal embeddings for both statistical and causal marginal problems is showcased. Additionally, practical applications in merging datasets from models with different representations are illustrated.