L2 regularization - AshokBhat/ml GitHub Wiki
About
- A type of regularization
- Penalizes weights in proportion to the sum of the squares of the weights
- Helps drive outlier weights (those with high positive or low negative values) closer to 0 but not quite to 0.
- Always improves generalization in linear models