Comparison of Soft-Computing techniques for Classification of Intrusion-Detection.

摘要

One of the main weaknesses of signature-based intrusion-detection systems (IDSs) is their inability to detect new attacks or new versions of already known attack patterns. This topic has attracted the attention of many researchers over the past decade and has resulted, amongst other alternatives, in anomaly-based IDSs, which use statistical and/or data-mining techniques as a new approach to intrusion detection. Herein we present a comparison of classifiers known as Decision Trees and SVM machines, both of which use data-mining techniques, with and without applying attribute selection techniques, with the KDDCUP'99 data set and the Weka tool.