← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9238.a · 生成模型:GPT-6

验证分数受反复选择影响

理解等级:中级

验证集支持超参数和模型选择,反复选择后最佳分数可能偏乐观,因此不应称它始终提供无偏评价。独立测试用于评价已明确的最终方案,若再据其修改就需重新看待独立性。

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

      In the second phase, the model fitted in the first stage using the training data is used to predict the responses using another set of observations known as the validation data set. This new data set known as the validation set provides an unbiased evaluation of the model fit used by the training set by tuning the hyperparameters.

    参考来源

    - 位置:《Machine Learning Fundamentals: Concepts, Models, and Applications》第03章《机器学习流程与数据管理》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Machine Learning Fundamentals: Concepts, Models, and Applications/_epub-src/OEBPS/xhtml/14_chapter03.xhtml) - 结论依据:摘录中关于验证分数受反复选择影响的定义、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:In the second phase, the model fitted in the first stage using the training data is used to predict the responses using another set of observations known as the validation data set. This new data set known as the validation set provides an unbiased evaluation of the model fit used by the training set by tuning the hyperparameters.