Application of Feature Selection and Fuzzy ARTMAP to Intrusion Detection

Author

Vilakazi, Christina B. ; Marwala, Tshilidzi

Author_Institution

Witwatersrand Univ., Johannesburg

Volume

6

fYear

2006

fDate

8-11 Oct. 2006

Firstpage

4880

Lastpage

4885

Abstract

This paper proposes a novel approach for intrusion detection and diagnosis. The proposed approach uses Sequential Backward Floating Search for feature selection and fuzzy ARTMAP for detection and diagnosis of attacks. The optimal vigilance parameter for the fuzzy ARTMAP is chosen using a genetic algorithm. The reduced set of features decreases the computation time by 0.789 s. A classification rate of 100% and 99.89% is obtained for the detection stage and diagnosis stage, respectively.

Keywords

ART neural nets; computer networks; feature extraction; fuzzy neural nets; genetic algorithms; pattern classification; search problems; security of data; telecommunication security; attack diagnosis; computer network; feature selection; fuzzy ARTMAP; genetic algorithm; intrusion detection; pattern classification; sequential backward floating search; Africa; Classification tree analysis; Computer networks; Cybernetics; Data mining; Decision trees; Educational institutions; Genetic algorithms; Intrusion detection; Protection;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems, Man and Cybernetics, 2006. SMC '06. IEEE International Conference on

Conference_Location

Taipei

Print_ISBN

1-4244-0099-6

Electronic_ISBN

1-4244-0100-3

Type

conf

DOI

10.1109/ICSMC.2006.385078

Filename

4274687