知识卡片
岭惩罚须区分范数与平方
通常岭回归使用λ‖θ‖₂²,而非仅λ‖θ‖₂;二者的梯度和解不同。原书一般式写Lp范数,后续岭例又明确用平方,阅读时必须保留这个区别。参数尺度与特征尺度也会影响惩罚含义,不能仅凭“L2”名称判断实现一致。
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
An especially common choice is p = 2, in which case the regularizer is called ridge regularization (which is also known as Tikonov regression). In the context of neural networks, this regularizer is called weight decay. When p = 1, the regularizer, applied to regression problems, is called LASSO regression. For any p, the effect is to encourage most parameters to be zero, or near
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:摘录中关于岭惩罚须区分范数与平方的定义、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:An especially common choice is p = 2, in which case the regularizer is called ridge regularization (which is also known as Tikonov regression). In the context of neural networks, this regularizer is called weight decay. When p = 1, the regularizer, applied to regression problems, is called LASSO regression. For any p, the effect is to encourage most parameters to be zero, or near