← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9383.d · 生成模型:GPT-6

交叉验证每折重新训练完整流程

理解等级:中级

将训练材料划为k折,每次用k−1折拟合,余下一折评价,再汇总结果。预处理也应在各折训练内拟合,避免泄漏;通常按选定超参重训最终模型,而不是直接挑最高分那折,另留测试。

这张卡片看不懂?

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

      Another approach to the train/test split question is cross-validation or k-fold cross-validation (see figure D.4). The concept behind cross-validation is very similar to the rough splits we just covered, but it allows you to use the entire labeled set as training. The process involves dividing your training set into k equal sets, or folds. You then train your model with k – 1 of the

    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:摘录中关于交叉验证每折重新训练完整流程的定义、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:Another approach to the train/test split question is cross-validation or k-fold cross-validation (see figure D.4). The concept behind cross-validation is very similar to the rough splits we just covered, but it allows you to use the entire labeled set as training. The process involves dividing your training set into k equal sets, or folds. You then train your model with k – 1 of the