Material covered
- Equivalence between maxent and LR classifiers
- Loss function for LR/maxent classifier compared to other loss functions
- Back to the exact hinge loss
- Gearing up to solve hinge-loss SVM
- Duality theory recap: Gordan's theorem, dual feasibility, complementary slackness, LP example
|
Last modified: Friday, 31 August 2007, 10:06 AM