Constrained Oversampling: An Oversampling Approach to Reduce Noise Generation in Imbalanced Datasets With Class Overlapping.
Changhui LiuSun JinDonghong WangZichao LuoJianbo YuBinghai ZhouChanglin YangPublished in: IEEE Access (2022)
Keyphrases
- imbalanced datasets
- class imbalance
- minority class
- majority class
- class noise
- class distribution
- cost sensitive learning
- learning from imbalanced data
- active learning
- sampling methods
- cost sensitive
- imbalanced data
- classification error
- decision boundary
- binary classification
- ensemble learning
- training set
- missing data
- support vector machine
- training dataset
- concept drift
- ensemble methods
- multi class
- nearest neighbour
- decision trees
- original data
- misclassification costs
- learning algorithm
- data sets