Ridge (statistics)

Regularization method that tends to shrink all coefficients to non-zero values

Related

• Lasso (statistics)
• Elastic Nets

$\lambda \sum _{j=1}^J{\beta }_j^2=\lambda \Vert \beta {\Vert }_2$

• This ridge term effectively minimizes the coefficients for any feature $\beta$
• $J$ is the number of parameters
• Effectively uses the L2 norm on each parameter
• This term is added to the loss function