Cross-validation and bootstrapping are unreliable in small sample classification.
Anders IsakssonMikael WallmanHanna GöranssonMats G. GustafssonPublished in: Pattern Recognit. Lett. (2008)
Keyphrases
- cross validation
- small sample
- model selection
- ls svm
- sample size
- support vector
- nearest neighbor classifiers
- least squares support vector machine
- hyperparameters
- cross validated
- training set
- variable selection
- support vector machine
- feature selection
- high dimensionality
- regression model
- classification accuracy
- machine learning
- parameter estimation
- feature vectors
- pattern recognition
- feature extraction