Stochastic Marginal Likelihood Gradients using Neural Tangent Kernels.
Alexander ImmerTycho F. A. van der OuderaaMark van der WilkGunnar RätschBernhard SchölkopfPublished in: CoRR (2023)
Keyphrases
- marginal likelihood
- model selection
- gaussian process
- closed form
- information criterion
- gaussian processes
- hyperparameters
- neural network
- exponential family
- feature space
- kernel function
- monte carlo
- support vector
- bayesian information criterion
- multiple kernel learning
- approximate inference
- kernel methods
- bayesian framework
- prior information
- maximum a posteriori