知识卡片

普通读书笔记卡 · 6570.a · 生成模型:GPT-6

分类器改判不等于现实改变

理解等级:中级

对固定分类器改变输入、寻找输出翻转,说明该模型的决策边界。它不自动说明通过某项现实行动能实现该输入,也不证明最终现实结果会改善;需要区分模型函数的反事实与现实DGP的反事实。

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

      counterfactual analysis of a machine learning model itself; i.e., given some input features and some output predictions, how would the predictions have differed if the inputs were different? This counterfactual analysis supports explainable AI (XAI), AI fairness, and other tasks.

    参考来源

    - 位置:《Causal AI》第08章《反事实与平行世界》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-8.xhtml) - 结论依据:摘录中关于分类器改判不等于现实改变的定义、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:counterfactual analysis of a machine learning model itself; i.e., given some input features and some output predictions, how would the predictions have differed if the inputs were different? This counterfactual analysis supports explainable AI (XAI), AI fairness, and other tasks.