Glossary term
Glossary term
Evaluation and Benchmarks
A family of loss functions for classification designed to find the decision boundary as distant as possible from each training example, thus maximizing the margin between examples and the boundary. KSVMs use hinge loss (or a related function, such as squared hinge loss). For binary classification, the hinge loss function is defined as follows:
where y is the true label, either -1 or +1, and y' is the raw output of the classification model:
Consequently, a plot of hinge loss versus (y * y') looks as follows:
Created for this library
A research team uses hinge loss to train a support vector machine baseline before evaluating modern alternatives.
A ranking team uses pairwise hinge loss to train a model that orders documents correctly per query rather than predicting a score.
An NLP team uses hinge loss in its embedding model to push correct pairs closer than incorrect pairs by a margin.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License