知识卡片

普通读书笔记卡 · 8836.a · 生成模型:GPT-6

差分隐私限制参与导致的变化

理解等级:高级

差分隐私用概率约束限制:相邻数据集只差一位参与者时,机制输出分布能改变多少。它提供关于额外泄露的数学边界,而不是保证一个人绝不被推断、绝无伤害。噪声、参数、邻接定义和重复发布都影响保证,不能把“用了随机性”直接等同满足差分隐私。

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

      A very successful theoretical privacy framework, called differential privacy, addresses these issues. Following the techniques of differential privacy [111] researchers can guarantee to the participants of a study that they will not be affected by allowing their data to be used in a study or analysis, no matter what other studies, datasets, or information sources are available.

    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:摘录中关于差分隐私限制参与导致的变化的定义、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:A very successful theoretical privacy framework, called differential privacy, addresses these issues. Following the techniques of differential privacy [111] researchers can guarantee to the participants of a study that they will not be affected by allowing their data to be used in a study or analysis, no matter what other studies, datasets, or information sources are available.