Network intrusion detection based on neuro-fuzzy classification

With rapid growth of computer networks during the past few years, network security has become a crucial issue. Among the various network security measures, intrusion detection systems (IDS) play a vital role to integrity, confidentiality and availability of resources. It seems that the presence of uncertainty and the imprecise nature of the intrusions make fuzzy systems suitable for such systems. Fuzzy systems are not normally adaptive and have not the ability to construct models solely based on the target system's sample data. One of the successful approaches which are incorporated fuzzy systems with adaptation and learning capabilities is the neural fuzzy method. The main objective of this work is to utilize ANFIS (adaptive neuro fuzzy inference system) as a classifier to detect intrusions in computer networks. This paper evaluates performance of ANFIS in the forms of binary and multi-classifier to categorize activities of a system into normal and suspicious or intrusive activities. Experiments for evaluation of the classifiers were performed with the KDD Cup 99 intrusion detection dataset. The Overall Results show that ANFIS can be effective in detecting various intrusions.