Computing functions of random variables via reproducing kernel Hilbert space representations.
Bernhard SchölkopfKrikamol MuandetKenji FukumizuStefan HarmelingJonas PetersPublished in: Stat. Comput. (2015)
Keyphrases
- random variables
- reproducing kernel hilbert space
- graphical models
- probability distribution
- loss function
- kernel methods
- bayesian networks
- special case
- latent variables
- euclidean space
- distance measure
- learning theory
- domain adaptation
- kernel function
- gaussian process
- real valued
- density estimation
- learning problems
- linear model
- data dependent
- marginal distributions
- shape analysis
- metric learning
- support vector machine
- probabilistic model
- positive definite