Training of intelligent intrusion detection system using neuro fuzzy

Abstract

Intrusion detection systems classify computer activities into two main categories: normal and suspicious activities. In order to achieve the classification, Intrusion detection systems use software computing techniques including neural networks and neuro fuzzy networks to categorize network activities and specify what category of attack is being generated. Neuro-Fuzzy classifiers are used for the initial classification of the initial network traffic. An inference system, Fuzzy inference systems is further used to determine whether the activity is normal or malicious. Efficient IDS systems are those capable of reducing false positives and generate high rate attack detection. However, fuzzy inference systems use human knowledge to create their fuzzy rule. In order to introduce a more accurate way of classifying network traffic, we introduce the use of Genetic Algorithms in conjunction with ANFIS so as to optimize data classification and obtain the best results. Genetic algorithms use a set of genetic operators such as mutation, crossover and selection on current population to reproduce similar patterns that will be used repeatedly until a particular criterion is met.