The Hidden Cost of Fraud: An Instance-Dependent Cost-Sensitive Approach for Positive and Unlabeled Learning.
Carlos Ortega VázquezJochen De WeerdtSeppe vanden BrouckePublished in: LIDTA (2022)
Keyphrases
- cost sensitive
- active learning
- fraud detection
- cost sensitive classification
- cost sensitive learning
- multi class
- supervised learning
- learning algorithm
- misclassification costs
- unsupervised learning
- class imbalance
- directly optimize
- data mining
- binary classification
- learning tasks
- learning process
- reinforcement learning
- inductive inference
- experimental design
- labeled data
- feature space
- decision trees