Sparsity of SVMs that use the epsilon-insensitive loss.
Ingo SteinwartAndreas ChristmannPublished in: NIPS (2008)
Keyphrases
- support vector
- hinge loss
- multi class
- support vector machine svm
- high dimensional
- kernel function
- machine learning
- feature selection
- loss function
- svm classifier
- convex loss functions
- maximum margin
- classification using support vector machines
- training support vector machines
- sparse kernel
- risk minimization
- cost sensitive
- statistical learning theory
- sparse representation
- generalization ability
- binary classification
- support vectors
- hyperplane
- reproducing kernel hilbert space
- structured learning
- linear support vector machines
- soft margin
- support vector machine classifiers
- model selection
- training data
- face recognition
- neural network