On the Potential Inadequacy of Mutual Information for Feature Selection: Proceedings of the 20th International Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2012)

Benoît Frénay, Gauthier Doquire, Michel Verleysen

Research output: Contribution in Book/Catalog/Report/Conference proceedingConference contribution

Abstract

Despite its popularity as a relevance criterion for feature selection, the mutual information can sometimes be inadequate for this task. Indeed, it is commonly accepted that a set of features maximising the mutual information with the target vector leads to a lower probability of misclassification. However, this assumption is in general not true. Justifications and illustrations of this fact are given in this paper.
Original languageEnglish
Title of host publicationProceedings of the 20th International Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2012)
Publisheri6doc.com
Publication statusPublished - 2012
Externally publishedYes

Keywords

  • ICTEAM:MLAI

Fingerprint Dive into the research topics of 'On the Potential Inadequacy of Mutual Information for Feature Selection: Proceedings of the 20th International Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2012)'. Together they form a unique fingerprint.

  • Cite this

    Frénay, B., Doquire, G., & Verleysen, M. (2012). On the Potential Inadequacy of Mutual Information for Feature Selection: Proceedings of the 20th International Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2012). In Proceedings of the 20th International Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning (ESANN 2012) i6doc.com.