Research on Neuro-fuzzy Inference System in Hierarchical Intrusion Detection

Intrusion detection technique has become increasingly important in the area of network security research. It is innovative that various soft computing approaches have been applied to the intrusion detection field. This paper presents an intelligent intrusion detection system which incorporates several soft computing techniques to implement either misuse or anomaly detection. Genetic algorithm is used to optimize the structure of the system. In the proposed system principal component analysis neural network is used to reduce the dimensions of the feature space. An enhanced fuzzy c-means clustering algorithm is used to cluster the preprocessed data to obtain fuzzy rules. And a hierarchical neuro-fuzzy classifier is developed. The experiments and evaluations of the proposed method were performed with the KDD Cup 99 intrusion detection dataset. Results indicate the high detection accuracy for intrusion attacks and low false alarm rate of the reliable system.