Non-enumerative Cross Validation for the Determination of Structural Parameters in Feature-Selective SVMs.
Elena ChernousovaPavel LevdikAlexander TatarchukVadim MottlDavid WindridgePublished in: ICPR (2014)
Keyphrases
- cross validation
- support vector
- hyperparameters
- model selection
- regularization parameter
- regression problems
- training set
- ls svm
- variable selection
- grid search
- unseen data
- kernel parameters
- cross validated
- svm classification
- selected features
- error estimates
- support vector machine
- rbf kernel
- generalization error
- svm classifier
- generalization ability
- feature vectors
- logistic regression
- feature selection
- input variables
- binary classification
- gaussian process
- parameter estimation
- feature set
- information criterion
- nearest neighbor classifiers
- parameter space
- data sets
- classification accuracy
- kernel function
- kernel methods
- radial basis function