Cost-sensitive elimination of mislabeled training data.
Donghai GuanWeiwei YuanTinghuai MaAsad Masood KhattakFrancis ChowPublished in: Inf. Sci. (2017)
Keyphrases
- cost sensitive
- training data
- naive bayes
- class distribution
- misclassification costs
- multi class
- support vector machine
- cost sensitive classification
- cost sensitive learning
- active learning
- classification accuracy
- test data
- binary classification
- decision trees
- data sets
- class imbalance
- training set
- supervised learning
- learning algorithm
- classification models
- training examples
- confidence weighted
- test set
- probability estimation
- generalization error
- prior knowledge
- feature vectors
- classification trees
- experimental design
- error rate
- base classifiers
- class labels
- test cases
- training samples
- image classification
- knn
- machine learning