Information-Theoretic Representation Learning for Positive-Unlabeled Classification.
Tomoya SakaiGang NiuMasashi SugiyamaPublished in: Neural Comput. (2021)
Keyphrases
- information theoretic
- supervised learning
- positive and unlabeled examples
- information theory
- mutual information
- information bottleneck
- unsupervised learning
- active learning
- prior knowledge
- learning algorithm
- pattern recognition
- image classification
- relative entropy
- decision trees
- information theoretic measures
- theoretic framework
- multi modality
- log likelihood
- class labels
- entropy measure
- labeled data
- model selection
- feature vectors