Theoretical and empirical study on the potential inadequacy of mutual information for feature selection in classification.
Benoît FrénayGauthier DoquireMichel VerleysenPublished in: Neurocomputing (2013)
Keyphrases
- mutual information
- feature selection
- conditional mutual information
- classification accuracy
- text classification
- classification models
- support vector
- image registration
- information gain
- feature extraction
- theoretical and empirical study
- information theoretic
- text categorization
- support vector machine
- feature set
- high dimensionality
- machine learning
- method for feature selection
- feature selection algorithms
- feature space
- feature subset selection
- decision trees
- feature subset
- irrelevant features
- selection criterion
- mutual information maximization
- classification algorithm
- pattern recognition
- sequential forward
- feature selection and classification
- training set
- classification performances
- discriminative features
- dimensionality reduction
- supervised learning
- cost function
- support vector machine svm
- decision making
- multi objective
- feature reduction
- computational intelligence
- feature weighting
- ensemble classifier
- model selection
- similarity measure
- knn
- data mining
- class labels
- naive bayes