Novel Mahalanobis-based feature selection improves one-class classification of early hepatocellular carcinoma.
Ricardo de Lima ThomazPedro C. CarneiroJoão Eliton BoninTúlio Augusto Alves MacedoAna Claudia PatrocinioAlcimar Barbosa SoaresPublished in: Medical Biol. Eng. Comput. (2018)
Keyphrases
- feature selection
- text categorization
- supervised and unsupervised learning
- feature set
- feature selection algorithms
- support vector
- novelty detection
- feature space
- mutual information
- multi class
- text classification
- machine learning
- selected features
- feature weighting
- classification models
- high dimensionality
- small sample
- unsupervised learning
- mahalanobis distance
- imbalanced data
- classification accuracy
- dimensionality reduction
- fold cross validation
- feature extraction
- density estimation
- risk factors
- high dimensional
- metric learning
- distance metric
- model selection
- distance measure
- nearest neighbor