Diversity sampling is an implicit regularization for kernel methods.
Michaël FanuelJoachim SchreursJohan A. K. SuykensPublished in: CoRR (2020)
Keyphrases
- kernel methods
- reproducing kernel hilbert space
- kernel ridge regression
- kernel matrices
- kernel function
- reproducing kernel
- learning problems
- support vector
- svm training
- machine learning
- kernel matrix
- feature space
- rademacher complexity
- support vector machine
- learning tasks
- random sampling
- multiple kernel
- kernel machines
- kernel learning
- graph kernels
- positive definite
- distance measure
- feature selection
- tikhonov regularization
- kernel trick
- kernel principal component analysis
- high dimensional feature space
- training samples
- multiple kernel learning
- loss function
- data dependent
- regularization parameter
- sample size