← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6578.b · 生成模型:GPT-6

模拟反事实要共享个体背景

理解等级:中级

模拟器要生成同一单位的反事实配对,应固定其初始状态、外生随机量和未干预机制,再更换目标机制。重复运行但重新抽全部噪声,只得到另一个样本;模拟器中的有效配对也不自动证明现实世界机制准确。

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

      various constraints. For one example with a given allocation outcome in the log, we could generate a counterfactual outcome for that example by applying a different allocation policy to that example. Similarly, given data produced by simulation software, we could generate counterfactual data by changing the simulation to reflect a hypothetical condition and then rerunning it with the same initial conditions as the original data.

    参考来源

    - 位置:《Causal AI》第10章《识别与因果层级》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-10.xhtml) - 结论依据:摘录中关于模拟反事实要共享个体背景的定义、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:various constraints. For one example with a given allocation outcome in the log, we could generate a counterfactual outcome for that example by applying a different allocation policy to that example. Similarly, given data produced by simulation software, we could generate counterfactual data by changing the simulation to reflect a hypothetical condition and then rerunning it with the same initial conditions as the original data.