Emerging Technologies for Developing Countries. First International EAI Conference, AFRICATEK 2017, Marrakech, Morocco, March 27-28, 2017 Proceedings

Research Article

Intrusion Detection Using Unsupervised Approach

  • @INPROCEEDINGS{10.1007/978-3-319-67837-5_18,
        author={Jai Singh and Nizar Bouguila},
        title={Intrusion Detection Using Unsupervised Approach},
        proceedings={Emerging Technologies for Developing Countries. First International EAI Conference, AFRICATEK 2017, Marrakech, Morocco, March 27-28, 2017 Proceedings},
        proceedings_a={AFRICATEK},
        year={2017},
        month={10},
        keywords={Mixture models Intrusion Aitchison distance Feature selection},
        doi={10.1007/978-3-319-67837-5_18}
    }
    
  • Jai Singh
    Nizar Bouguila
    Year: 2017
    Intrusion Detection Using Unsupervised Approach
    AFRICATEK
    Springer
    DOI: 10.1007/978-3-319-67837-5_18
Jai Singh1,*, Nizar Bouguila1,*
  • 1: Concordia University
*Contact email: jaipuneet.singh@mail.concordia.ca, nizar.bouguila@concordia.ca

Abstract

The process of detecting intrusion on network traffic has always remained a key concern for security researchers. During the previous years, intrusion detection had attracted many researchers to find anomaly on NSL-KDD data set. Hence, most of the approaches applied on NSL KDD data set were supervised approaches. We had conducted statistical analysis on this data set using Dirichlet Mixture model. We have seen initialization using Aitchison distance fits better for proportional data. The feature selection highly affects both the performance and results into an improved evaluation of anomaly detection by an unsupervised approach.