$\color{black}\rule{365px}{3px}$

http://image.diku.dk/imagecanon/material/cortes_vapnik95.pdf

Loss Function (Hinge Loss)

$\color{black}\rule{365px}{3px}$

Hinge loss is designed to maximize the margin between data points and the decision boundary (hyperplane). It not only wants correct classification but also pushes correct predictions far away from the decision boundary.

For a given training example $(x_i, y_i)$, where:

• $x_i$ is the input feature.
• $y_i$ is the true label (either $+1$ or $−1$ for binary classification).