Efficient optimization of hyper-parameters for least squares support vector regression.
Andreas FischerGerd LangensiepenKlaus LuigNico StrasdatThorsten ThiesPublished in: Optim. Methods Softw. (2015)
Keyphrases
- efficient optimization
- hyperparameters
- model selection
- cross validation
- closed form
- bayesian framework
- random sampling
- bayesian inference
- prior information
- support vector
- noise level
- em algorithm
- maximum a posteriori
- sample size
- incremental learning
- incomplete data
- optimization methods
- maximum likelihood
- missing values
- convex relaxation
- parameter space
- ls svm
- image processing
- high quality
- active learning
- parameter settings
- computer vision
- expectation maximization