知识卡片

普通读书笔记卡 · 6685.c · 生成模型:GPT-6

多项式特征需按复杂度与目标选择

理解等级:中级

加入平方和交互项可让线性模型表达非线性关系,但项数随特征数及阶数迅速增加,估计和验证成本也增加。不能由此推出三变量乘积物理上不可能,或所有任务最高只能二阶;应按问题结构、正则与验证选择。

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 example of some feature engineering that neural networks can optimize for you is polynomial feature extraction (think back to the last time you used sklearn .preprocessing.PolynomialFeatures). During feature engineering, you might guess that the relationship between inputs and outputs is quadratic. In that case, you would square those input features and retrain a model with these new features to see if it improved your

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第05章《神经网络基础》(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-5.html) - 结论依据:摘录中关于多项式特征需按复杂度与目标选择的定义、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 example of some feature engineering that neural networks can optimize for you is polynomial feature extraction (think back to the last time you used sklearn .preprocessing.PolynomialFeatures). During feature engineering, you might guess that the relationship between inputs and outputs is quadratic. In that case, you would square those input features and retrain a model with these new features to see if it improved your