知识卡片
错误代价应进入目标和决策
先与消费方明确不同错误的实际后果,再选数据权重、损失、评价或推理阈值处理。按类别频数加权F1不自动等于业务成本加权,重采样也不保证修复所有错误与概率校准。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 模型自动化测试
原书出处 · 1 处
- · 查看原书依据
cost-sensitive learning techniques to train or condition your models to be more cautious in high-stakes scenarios and be less so where the consequences of errors are minimal. You also need to collaborate with the downstream consumers of your ML system to understand the cost of different types of failures.
参考来源
- 位置:《Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)》第06章《模型自动化测试》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Effective Machine Learning Teams (David Tan, Ada Leung, David Colls)/_epub-src/OEBPS/ch06.html)
- 结论依据:摘录中关于错误代价应进入目标和决策的定义、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:cost-sensitive learning techniques to train or condition your models to be more cautious in high-stakes scenarios and be less so where the consequences of errors are minimal. You also need to collaborate with the downstream consumers of your ML system to understand the cost of different types of failures.