Intrusion Detection Mechanism Based On Modular Neural Network

摘要

Nowadays, the complexity and diversity of network security issues have caused intrusion detection to become an important research topic. Aiming at the problems of traditional intrusion detection mechanisms, the modular neural network intrusion detection mechanism based on K-means similarity judgment is proposed. The method uses the Meanshift-Kmeans clustering algorithm to calculate the clustering center and data classification, then uses the similarity to determine the dynamic allocation of sub-modules, and finally uses the classifiers trained according to different types of intrusion to achieve intrusion detection. Through experiments on two different intrusion data sets, the experimental results prove the good detection efficiency and model stability of the model.