知识卡片

普通读书笔记卡 · 8982.b · 生成模型:GPT-6

密度归一化不等于点概率

理解等级:中级

连续密度非负、全空间积分为1,密度数值可以大于1;某个区域的概率需要积分。提高某些区域的质量必须从其他区域转移,但有限样本点上的密度与其精确命中概率不同。比较密度还需相同变量、单位及基准测度。

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

      Remember that a probability density function (pdf) is a function pθ : X → [0, ∞) with ∫x pθ(x)dx = 1 (i.e. it's normalized). To learn a pdf, we will typically learn the parameters of a family of pdfs. For example, in section 32.6, we will cover learning the parameters of the Gaussian family of pdfs. All members of such a the family are normalized

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第32章《生成模型》(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/c032.xhtml) - 结论依据:摘录中关于密度归一化不等于点概率的定义、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:Remember that a probability density function (pdf) is a function pθ : X → [0, ∞) with ∫x pθ(x)dx = 1 (i.e. it's normalized). To learn a pdf, we will typically learn the parameters of a family of pdfs. For example, in section 32.6, we will cover learning the parameters of the Gaussian family of pdfs. All members of such a the family are normalized