知识卡片

普通读书笔记卡 · 6560.h · 生成模型:GPT-6

结构差异应按目标查询比较

理解等级:高级

两个SCM可能观测等价,却在某些干预或反事实查询上不同,也可能只是外生坐标的等价重写。比较时先规定保持的观测量、允许的干预及跨情境规则,再判断目标查询是否相同,而不是只比较公式外观。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      equivalence classes of SCMs—given a causal DAG and a joint distribution, there can be multiple SCMs consistent with that DAG and distribution. We saw this with how the two variants of the stick-breaking die-roll SCM are both consistent with the DAG X (die choice) → Y (die roll) and with the distributions P (X ) (probability distribution on die selection) and P (Y |X )

    参考来源

    - 位置:《Causal AI》第06章《结构因果模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-6.xhtml) - 结论依据:摘录中关于结构差异应按目标查询比较的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:equivalence classes of SCMs—given a causal DAG and a joint distribution, there can be multiple SCMs consistent with that DAG and distribution. We saw this with how the two variants of the stick-breaking die-roll SCM are both consistent with the DAG X (die choice) → Y (die roll) and with the distributions P (X ) (probability distribution on die selection) and P (Y |X )