How to evaluate an agent's behavior to infrequent events? - Reliable performance estimation insensitive to class distribution.
Sirko StraubeMario Michael KrellPublished in: Frontiers Comput. Neurosci. (2014)
Keyphrases
- class distribution
- training data
- rare events
- cost sensitive
- training set
- highly skewed
- class imbalance
- training samples
- test set
- imbalanced data
- estimation error
- imbalanced datasets
- training set size
- machine learning methods
- cost sensitive learning
- class labels
- majority class
- base classifiers
- unlabeled data
- text mining
- highly imbalanced