A Novel Information Theoretic Objective to Disentangle Representations for Fair Classification.
Pierre ColomboNathan NoiryGuillaume StaermanPablo PiantanidaPublished in: IJCNLP (Findings) (2023)
Keyphrases
- information theoretic
- mutual information
- information theory
- theoretic framework
- pattern recognition
- entropy measure
- decision trees
- multi modality
- information bottleneck
- feature vectors
- text classification
- classification accuracy
- machine learning
- feature selection
- support vector
- information theoretic measures
- log likelihood
- relative entropy
- minimum description length
- multi modal
- kullback leibler divergence
- density estimation
- supervised learning
- feature space
- learning algorithm