Improved Cross-Validation for Classifiers that Make Algorithmic Choices to Minimise Runtime Without Compromising Output Correctness.
Dorian FlorescuMatthew EnglandPublished in: MACIS (2019)
Keyphrases
- cross validation
- support vector
- training set
- nearest neighbor classifiers
- unseen data
- model selection
- binary classifiers
- variable selection
- hyperparameters
- roc analysis
- grid search
- generalization error
- feature selection
- cross validated
- svm classifier
- decision trees
- training data
- test set
- improved algorithm
- input variables
- regularization parameter
- leave one out cross validation
- information criterion
- support vector machine
- multiclass classification
- nearest neighbor
- data sets
- error estimates
- small sample
- binary classification
- roc curve
- machine learning algorithms
- multi class
- neural network