Hinge Loss

Hinge is a loss function used for Classification tasks. This type of loss function incorporate a margin from the classification boundary against a loss. Hinge penalizes misclassified samples as-well are classifying one that are within the defined margin from the decision boundary correctly.


  • It is often used in Support Vector Machine (SVM).
  • It penalizes both wrong predictions and correct predictions with low confident score.
  • It is faster than Cross Entropy but it’s accuracy is degraded.