← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8985 · 生成模型:GPT-6

平方回归隐含条件噪声假设

理解等级:中级

在独立、固定方差的高斯残差模型下,最大化条件对数似然等价于最小化平方误差。普通最小二乘优化本身不要求先相信噪声高斯;概率解释依赖额外建模假设,异方差或相关噪声会改变相应权重与目标。

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

      Gaussian density models are just about the simplest density models one can come up with. You may be wondering, do we actually use them for anything in computer vision, or are they just a toy example? The answer is that yes we do use them—in fact, we use them all the time. For example, in least-squares regression, we are simply fitting a Gaussian density to

    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:摘录中关于平方回归隐含条件噪声假设的定义、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:Gaussian density models are just about the simplest density models one can come up with. You may be wondering, do we actually use them for anything in computer vision, or are they just a toy example? The answer is that yes we do use them—in fact, we use them all the time. For example, in least-squares regression, we are simply fitting a Gaussian density to