知识卡片
交叉验证估计而非消除过拟合
过拟合是模型过度适应训练样本而在新样本上表现下降。交叉验证通过轮换训练与验证子集估计表现,不会自动防止模型或调参过程过拟合。若反复根据同一验证结果选择方案,仍需独立验收。
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
Overfitting means the classifying model is too biased towards the training data and may not work well on unknown text data. Overfitting can be avoided by splitting labeled data into k-folds for training and testing purposes. It ensures testing is done on different data than trained with maximum utilization of test data.
参考来源
- 位置:《Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications》第02章《文本挖掘与垃圾邮件过滤》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications/_epub-src/OEBPS/Text/9789815136746-C2.html)
- 结论依据:摘录中关于交叉验证估计而非消除过拟合的定义、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:Overfitting means the classifying model is too biased towards the training data and may not work well on unknown text data. Overfitting can be avoided by splitting labeled data into k-folds for training and testing purposes. It ensures testing is done on different data than trained with maximum utilization of test data.