Kernel Choice and Classifiability for RKHS Embeddings of Probability Distributions.
Bharath K. SriperumbudurKenji FukumizuArthur GrettonGert R. G. LanckrietBernhard SchölkopfPublished in: NIPS (2009)
Keyphrases
- reproducing kernel hilbert space
- probability distribution
- kernel methods
- euclidean space
- hilbert space
- reproducing kernel
- kernel function
- distance measure
- loss function
- random variables
- regularized least squares
- density estimation
- kernel matrix
- positive definite
- vector space
- probability density function
- data dependent
- input space
- learning problems
- kernel machines
- learning theory
- dimensionality reduction
- feature space
- gaussian kernels
- real valued
- bayesian networks
- special case
- shape analysis
- domain adaptation
- optimal kernel
- gaussian process
- representer theorem
- learning tasks
- finite dimensional
- support vector
- training data
- von neumann
- graph kernels
- feature vectors
- multiple kernel learning
- data points
- metric learning
- machine learning