知识卡片
社会偏差横跨图与标签
地域覆盖、群体呈现与职业标签刻板关联,可能同时进入图像和标注。总体准确率不能充分显示不同人群的后果。应结合使用情境检查覆盖、标签假设与分组结果,并评估部署影响;书中医生与护士的性别关联是偏差示例,不是应沿用的分类规律。
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
This type of bias can affect both the images and the associated labels. Social bias refers to biases associated to how groups of people might be reflected differently in the data. Social biases can be originated by the data collection protocol or by existing social biases. Images can be captured in certain world regions, or contain stereotypes. Labels can reflect social assumptions and biases. An
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第35章《数据偏差与分布漂移》(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/c035b.xhtml)
- 结论依据:摘录中关于社会偏差横跨图与标签的定义、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:This type of bias can affect both the images and the associated labels. Social bias refers to biases associated to how groups of people might be reflected differently in the data. Social biases can be originated by the data collection protocol or by existing social biases. Images can be captured in certain world regions, or contain stereotypes. Labels can reflect social assumptions and biases. An