← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6573.a · 生成模型:GPT-6

增量均值不足以确定四类概率

理解等级:高级

某属性群体的平均增量等于受益型概率减去受害型概率。即使知道该差值,通常也不能分别恢复四种潜在响应类型的概率;把用户确定分到一类还需要跨世界假设与足够信息,普通分类训练不能直接观察两种结果标签。

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

      P(YX=1, YX=0|C=c). Various statistical segmentation methods seek to define C such that users fall into groups that have high probability for one of the four outcomes of P(YX=1, YX=0|C=c), but before we apply the stats, our first task will be to ensure we can estimate this query using sufficient assumptions and data. We’ll cover how to estimate counterfactuals with SCMs in chapter 9 and how

    参考来源

    - 位置:《Causal AI》第08章《反事实与平行世界》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-8.xhtml) - 结论依据:摘录中关于增量均值不足以确定四类概率的定义、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:P(YX=1, YX=0|C=c). Various statistical segmentation methods seek to define C such that users fall into groups that have high probability for one of the four outcomes of P(YX=1, YX=0|C=c), but before we apply the stats, our first task will be to ensure we can estimate this query using sufficient assumptions and data. We’ll cover how to estimate counterfactuals with SCMs in chapter 9 and how