Efficient information theoretic extraction of higher order features for improving neural network-based spam e-mail categorization.
Vasilios ZorkadisDimitris A. KarrasPublished in: J. Exp. Theor. Artif. Intell. (2006)
Keyphrases
- information theoretic
- higher order
- mutual information
- information theory
- theoretic framework
- entropy measure
- jensen shannon divergence
- neural network
- log likelihood
- feature vectors
- information bottleneck
- feature space
- information theoretic measures
- multi modality
- feature extraction
- minimum description length
- data mining
- natural images
- feature set
- information extraction
- pairwise
- search engine
- spam e mail