Support Vector Machine

Last updated Oct 28, 2022 Edit Source

• Seed
• CPSC340

An SVM is just Hinge loss with L2-regularization

$$f(w) = \sum_{i=1}^n \max{0, 1-y_iw^Tx_i}$$

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