← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 6595.c · 生成模型:GPT-6

最坏损失规则要明确最坏范围

理解等级:中级

最小最大损失在声明的结果、状态或模型集合上取最坏情况,再选择最小损失行动。它与按概率平均的期望效用不同;可行最坏集合若改变,建议也会改变,不能把极端罕见事件与模型不确定性不加区分地混合。

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

      Minimax decision rules To understand the minimax decision rule, recall that the terms “utility” and “loss” are two sides of the same coin; utility == negative loss. Let L(y ) = – U(y ). Then a minimax decision rule is In plain English, this means “choose the action that minimizes the maximum amount of possible loss.” In our investment example, this means choosing the investment

    参考来源

    - 位置:《Causal AI》第12章《因果决策与强化学习》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-12.xhtml) - 结论依据:摘录中关于最坏损失规则要明确最坏范围的定义、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:Minimax decision rules To understand the minimax decision rule, recall that the terms “utility” and “loss” are two sides of the same coin; utility == negative loss. Let L(y ) = – U(y ). Then a minimax decision rule is In plain English, this means “choose the action that minimizes the maximum amount of possible loss.” In our investment example, this means choosing the investment