知识卡片
数据歧义可用独立统计诊断
比较同一姓氏在多个来源标签中的出现情况,可揭示标注重叠或不平衡,无需先训练模型。用多数来源作参照的矩阵描述的是语料歧义,不是模型预测混淆;二者对象不同,不能用前者替代验证表现。
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
By computing the most popular nationality for each name in the dataset, it is possible to create a confusion matrix, using the most common nationality as the “true” label for a particular name. This can reveal several quirks in the dataset that should influence what the model learns and how well it can perform this task. There is no confusion at all for Arabic names
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第08章《循环网络与长短期记忆》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-8.html)
- 结论依据:摘录中关于数据歧义可用独立统计诊断的定义、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:By computing the most popular nationality for each name in the dataset, it is possible to create a confusion matrix, using the most common nationality as the “true” label for a particular name. This can reveal several quirks in the dataset that should influence what the model learns and how well it can perform this task. There is no confusion at all for Arabic names