知识卡片
数据选择与公平评价都需具体标准
采样、标注和特征会决定哪些情形被模型关注,并可能复制不平等。应明确受影响群体与误差,保存来源并检查分组结果;仅凭开发者觉得公平,或把训练集简单均衡,不足以证明部署无偏。
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
原书出处 · 1 处
- · 查看原书依据
D.1.3 Data selection and avoiding bias Data selection and feature engineering are ripe grounds for introducing bias (in human terms) into a model. Once you’ve baked your own biases into your algorithm by choosing a particular set of features, the model will fit to those biases and produce biased results. If you’re lucky enough to discover this bias before going to production, it can require
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》附录D《附录D:机器学习工具与技术》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/appendix-d.html)
- 结论依据:摘录中关于数据选择与公平评价都需具体标准的定义、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:D.1.3 Data selection and avoiding bias Data selection and feature engineering are ripe grounds for introducing bias (in human terms) into a model. Once you’ve baked your own biases into your algorithm by choosing a particular set of features, the model will fit to those biases and produce biased results. If you’re lucky enough to discover this bias before going to production, it can require