← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8963 · 生成模型:GPT-6

点估计取决于错误代价

理解等级:中级

同一个后验可以对应不同最佳决策。离散状态的等代价误分类损失导出后验最大概率状态;数值平方损失则导出后验均值。选择估计前先说明任务损失,不能脱离代价函数把MAP或均值称作普遍更好。

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

      Selecting the best estimate requires specifying a penalty for making a wrong guess. If we penalize all wrong answers equally, the best strategy is to guess the value of x that maximizes the posterior probability, p(x | y) (because any other explanation for the observations y would be less probable). That is called the maximum a posteriori (MAP) estimate.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第29章《概率图模型》(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/c029.xhtml) - 结论依据:摘录中关于点估计取决于错误代价的定义、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:Selecting the best estimate requires specifying a penalty for making a wrong guess. If we penalize all wrong answers equally, the best strategy is to guess the value of x that maximizes the posterior probability, p(x | y) (because any other explanation for the observations y would be less probable). That is called the maximum a posteriori (MAP) estimate.