Information-theoretic Classification Accuracy: A Criterion that Guides Data-driven Combination of Ambiguous Outcome Labels in Multi-class Classification.
Chihao ZhangYiling Elaine ChenShihua ZhangJingyi Jessica LiPublished in: J. Mach. Learn. Res. (2022)
Keyphrases
- cross validation
- information theoretic
- multi class classification
- support vector
- classification accuracy
- training set
- minimum description length
- binary classification
- mutual information
- feature selection
- information theory
- relative entropy
- multi class
- training examples
- generalization error
- hyperparameters
- support vector machine
- training data
- kernel function
- loss function
- computer vision
- naive bayes
- pairwise
- multi label
- multi task learning
- probabilistic model
- feature space
- high dimensional
- class labels
- text categorization