Low-cost model selection for SVMs using local features.
Miguel Lázaro-GredillaVanessa Gómez-VerdejoEmilio Parrado-HernándezPublished in: Eng. Appl. Artif. Intell. (2012)
Keyphrases
- model selection
- low cost
- cross validation
- feature selection
- machine learning
- support vector
- bayesian learning
- feature vectors
- statistical learning
- sample size
- parameter estimation
- hyperparameters
- feature space
- error estimation
- feature set
- statistical inference
- learning machines
- regression model
- generalization error
- svm classifier
- kernel function
- variable selection
- mixture model
- multi class
- feature extraction
- knn
- classification accuracy
- training set
- selection criterion
- objective function
- generalization bounds
- marginal likelihood
- model selection criteria