loss function

几种损失函数：
对于回归问题：
平方损失：[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=\(y-\hat{y}\)^2[/img]
绝对值损失：[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=|y-\hat{y}|[/img]
[img]http://latex.codecogs.com/gif.latex?\epsilon[/img]−insensitive损失：
[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=max\{|y-\hat{y}-\epsilon|,0\}=:|y-\hat{y}|_\epsilon[/img]
对于分类问题：
平方损失：[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=\(1-y\hat{y}\)^2[/img]
hinge loss:[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=max\{1-y\hat{y},0\}=:|y-\hat{y}|_+[/img]
logistic loss:[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=\(ln2\)^{-1}ln(1+e^{\(1-y\hat{y}\)}\)[/img]
Huber loss:[img]http://latex.codecogs.com/gif.latex?l\(y,\hat{y}\)=max\{1-y\hat{y},0\}^2[/img]