知识卡片
插值噪声不等于恢复规律
训练观测包含真实关系和随机误差。足够灵活的曲线能穿过每个观测点,也可能追随不在新数据中重复的噪声。零训练误差说明对当前样本的拟合,不说明已经找回生成规律;是否过拟合仍取决于未见数据误差,而非只看是否插值。
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
- · 泛化问题
原书出处 · 1 处
- · 查看原书依据
fit the noise, the minor offsets ϵ around the green line. This noise is a property of the training data that does not generalize to the test data; the test data will have different observation noise. That's what we mean when we say a model is overfitting.
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第11章《泛化问题》(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/c011.xhtml)
- 结论依据:摘录中关于插值噪声不等于恢复规律的定义、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:fit the noise, the minor offsets ϵ around the green line. This noise is a property of the training data that does not generalize to the test data; the test data will have different observation noise. That's what we mean when we say a model is overfitting.