On asymptotic properties of hyperparameter estimators for kernel-based regularization methods.
Bi-Qiang MuTianshi ChenLennart LjungPublished in: Autom. (2018)
Keyphrases
- asymptotic properties
- regularization methods
- regularization parameter
- hyperparameters
- total variation
- edge preserving
- support vector
- regularization method
- inverse problems
- fixed point
- gaussian processes
- image restoration
- model selection
- gaussian process
- kernel methods
- cross validation
- random sampling
- support vector machine
- noise reduction
- prior information
- denoising
- closed form
- structure tensor
- em algorithm
- bayesian inference
- global optimization
- iterative methods
- evolutionary algorithm
- sufficient conditions
- maximum a posteriori
- early vision
- image denoising
- mumford shah
- feature selection