← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6549.b · 生成模型:GPT-6

碰撞点需自身或后代被条件化

理解等级:中级

碰撞结构X→M←Y默认阻断该路径;若M或其任一后代进入条件集,该碰撞处被激活。但整条路径仍可能被其他非碰撞点或未激活碰撞点阻断,因此激活一个碰撞处不保证两端整体d连接。

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

      4. P contains a collider structure, i → m ← j, such that the middle node m is not in Z, and no descendant of m is in Z. This fourth criterion is how d-separation captures the way two independent (d-separated) items can become dependent when conditioning on a collider. Many writers conflate d-separation and conditional independence. Keep the distinction clear in your mind: ⊥G

    参考来源

    - 位置:《Causal AI》第04章《用因果约束检验DAG》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-4.xhtml) - 结论依据:摘录中关于碰撞点需自身或后代被条件化的定义、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:4. P contains a collider structure, i → m ← j, such that the middle node m is not in Z, and no descendant of m is in Z. This fourth criterion is how d-separation captures the way two independent (d-separated) items can become dependent when conditioning on a collider. Many writers conflate d-separation and conditional independence. Keep the distinction clear in your mind: ⊥G