Absolute integrability of Mercer kernels is only sufficient for RKHS stability.
Mauro BisiaccoGianluigi PillonettoPublished in: CoRR (2023)
Keyphrases
- mercer kernels
- gaussian kernels
- kernel methods
- reproducing kernel hilbert space
- loss function
- shape from shading
- feature space
- learning algorithm
- kernel function
- machine learning
- learning theory
- gaussian kernel
- support vector
- euclidean space
- data dependent
- random projections
- domain adaptation
- distance measure
- kernel learning
- learning rate
- gaussian process
- decision boundary
- kernel matrix
- multiple kernel learning