Smoothed Hinge Loss and ℓ1 Support Vector Machines.
Jeffrey HajewskiSuely OliveiraDavid E. StewartPublished in: ICDM Workshops (2018)
Keyphrases
- hinge loss
- loss function
- support vector
- soft margin
- learning machines
- binary classification
- risk minimization
- reproducing kernel hilbert space
- support vector machine
- generalization ability
- kernel function
- kernel methods
- multi class
- pairwise
- decision boundary
- potential functions
- support vectors
- classification accuracy
- parameter selection
- boosting algorithms
- training examples
- line search
- cross validation
- kernel matrix
- maximum margin
- pairwise constraints
- class conditional
- nearest neighbor