Intrusion Detection System Using Visualization and Integration Technique
No Thumbnail Available
Date
2006-08
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Intrusion detection is an area of ever increasing importance. Currently existing Intrusion
Detection Systems (IDS) lack visualization and false alarms detection capabilities.
Researchers have proposed integrated systems which may reduce the percentage of false
alarms. This work addresses the above stated problems by integrating Self-Organized Map
(SOM) with Genetic Algorithm (GA) so as to minimize the false alarms and also to provide
visualization capability to the new IDS. SOM is an unsupervised Artificial Neural Network
(ANN) learning algorithm that attempts to visualize a large dataset in compact representation.
GA is an evolutionary computing type of artificial intelligence algorithm, which is better for
optimization, feature selection and clustering problems. The performance of the model is
measured using Knowledge Discovery and Data Mining (KDD) Cup 99 dataset, which was
prepared for The Third International Knowledge Discovery and Data Mining (DM) Tools
Competition for researchers who work on intrusion detection. The work also includes GA
based feature selection to further improve the performance of the model. The result shows
94.3 % of intrusion detection rate with 2.93% of false alarm rate.
Description
Keywords
Computer Engineering