Application Of Several Data Mining Algorithms In Network Intrusion Detection

Network security problem is mainly caused by the lack of competent and Internet automatic defensive,prompted network attack behavior easy to invasion of the Internet.In order to further enhance its active defense capability,it is necessary to significantly improve its intrusion detection accuracy level,thereby significantly reducing the false alarm rate,while also weakening the problem of data overload,applying data mining technology to network intrusion detection,in order to achieve a has adaptability and good extension performance of intrusion detection system.Firstly,the combination of artificial immune algorithm and artificial immune algorithm network intrusion method,so as to achieve the purpose of protecting network security,show the efficiency of the algorithm.It simulates the principle of biological immunity and has the ability of distributed computing,self-monitoring and dynamic learning.The research results show that the detection accuracy of intrusion behavior and the reduction of false alarm rate have been improved,and the shortcomings of other network intrusion detection algorithms have been improved,which provides a new idea for network intrusion detection.Secondly,this paper establishes a Bayesian-based network intrusion detection model for simulation experiments.In this model,principal component analysis is used to extract key attributes of network data packets,eliminate redundant attributes,reduce dimensions,and then classify them with Bayesian classifier.The results show that the model achieves the required intrusion detection target,and the detection speed is accelerated,and the experimental results are satisfactory,but the deficiency of high false alarm rate still needs to be improved.The third approach is that the model of intrusion detection system is further innovated based on the optimized genetic algorithm neural network.That is to say,with the help of the genetic algorithm,the weights of the neural network(BP)are optimized so that they can be well integrated.At the same time,the trained BP neural network is used to detect the unmatched suspicious intrusion text,and the specific types of network intrusion can be identified.The simulation results in MATLAB show that the combination of genetic algorithm and improved BP neural network has great potential in the application of network intrusion detection.Compared with the traditional network intrusion detection system model,it has better intrusion recognition rate and detection effect.In addition to the above three methods,the deep learning algorithm is studied andanalyzed,and the convolutional neural model is established.The original data is transformed into higher and more abstract form through some simple non-linear models,which strengthens the ability to classify the original data and weakens the feature classification which is not related to the classification in the original data.Its greatest advantage is that it can accurately extract the local correlation of features and improve the accuracy of feature extraction.Finally,the simulation results of three kinds of network intrusion models based on data mining are compared,and the most effective network intrusion detection model is selected,which can effectively solve the problem of network intrusion detection behavior and improve the degree of automation.At the same time,it also strengthens the efficiency of detection preparation and self-adaptation.