Support Vector Machine

An SVM is just Hinge loss with L2-regularization

$f(w)={\sum }_{i=1}^n\max \{0,1-y_i{w}^T{x}_i\}$

They can also be viewed as ‘maximizing the margin’: