← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8836.b · 生成模型:GPT-6

随机回答需要校正总体比例

理解等级:高级

例中以1/2概率如实回答,以1/2概率再抛公平硬币随机答。若真实肯定率为p,观测肯定率期望q=1⁄4+p/2,所以估计应作p=2q−1/2;原书称观测百分比期望直接等于真实比例不准确。随机化掩护单次答案,但需更多样本且不表示完全没有隐私信息。

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

      Each surveyed adult is instructed to flip a coin. If the coin shows heads, they are instructed to answer the question, “Have you cheated on your partner?”, truthfully. If the coin shows tails, they are asked to answer “yes” or “no” at random, by flipping the coin again and answering “yes” if heads and “no” if tails. The resulting responses for each individual are stored.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第04章《计算机视觉与社会》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c004.xhtml) - 结论依据:摘录中关于随机回答需要校正总体比例的定义、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:Each surveyed adult is instructed to flip a coin. If the coin shows heads, they are instructed to answer the question, “Have you cheated on your partner?”, truthfully. If the coin shows tails, they are asked to answer “yes” or “no” at random, by flipping the coin again and answering “yes” if heads and “no” if tails. The resulting responses for each individual are stored.