A Convex Parametrization of a New Class of Universal Kernel Functions for use in Kernel Learning.
Brendon K. ColbertMatthew M. PeetPublished in: CoRR (2017)
Keyphrases
- kernel learning
- kernel function
- quadratically constrained quadratic
- kernel methods
- support vector
- optimal kernel
- support vector regression
- reproducing kernel hilbert space
- kernel matrix
- semi supervised
- positive semidefinite
- input space
- multiple kernel learning
- kernel trick
- kernel matrices
- gaussian kernels
- high dimensional feature space
- feature space
- support vector machine
- support vectors
- multiple kernel
- active learning
- multi task learning
- decision boundary
- class labels
- high dimensional
- convex sets
- hyperplane
- convex optimization
- pairwise constraints
- learning tasks
- feature set
- knn
- pairwise