知识卡片

普通读书笔记卡 · 9114.g · 生成模型:GPT-6

矩阵乘积与逐元素乘积不同

理解等级:初级

矩阵乘积AB按内维配对求和,点积是相应向量乘法的特例;Hadamard乘积A⊙B按对应元素相乘。二者输出、形状条件和语义不同,不能因编程语言某运算符写法简短就混用。本书又以专门符号区别卷积与互相关,阅读公式时先确认运算定义。

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

      Hadamard product (i.e., element-wise product): x ⊙ y, A ⊙ B Product of two scalars: ab or a * b In the geometry chapters, and in some other chapters, the variables x, y, z will denote location and will have a special treatment. Signal Processing Discrete signals (and images) are vectors x, y, z. When indexing a discrete signal we will use brackets: x [n],

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》符号约定专题《符号与矩阵微积分约定》(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/c000b.xhtml) - 结论依据:摘录中关于矩阵乘积与逐元素乘积不同的定义、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:Hadamard product (i.e., element-wise product): x ⊙ y, A ⊙ B Product of two scalars: ab or a * b In the geometry chapters, and in some other chapters, the variables x, y, z will denote location and will have a special treatment. Signal Processing Discrete signals (and images) are vectors x, y, z. When indexing a discrete signal we will use brackets: x [n],