svm-svr-epsilon-insensitive-loss

IN premise — entries/2026/06/21/wiki-Support_vector_machine-chunk-6.md

Created 2026-06-21T09:55:54+00:00

Support Vector Regression (SVR), introduced by Drucker et al. (1997), uses an epsilon-insensitive loss function.

Dependents

These beliefs depend on this one:

• IN svm-framework-extends-beyond-binary-classification — The SVM framework extends well beyond its original binary classification setting — SVR adapts the max-margin principle to regression via epsilon-insensitive loss, transductive SVMs bridge to semi-supervised learning by jointly optimizing over labeled and unlabeled data, and Bayesian SVMs reinterpret the framework probabilistically for automatic hyperparameter tuning with uncertainty quantification.