几种常用的loss function

---

0-1 Loss:      L01(mi(w))=I[mi(w)<0]
Log loss:      Llog(mi(w))=log(1+e−mi(w))
Hingle loss:   Lhingle(mi(w))=max(0,1−mi(w))

其中，w 表示weight vector，ti表示class label，xi表示feature vector，mi(w)=tifw(xi)。

Reference

[1]. Algorithms for Direct 0–1 Loss Optimization in Binary Classification