Research On Intrusion Detection Technology Based On Optimized Random Forest Algorithm

With the rapid development of the Internet in the world,the security of computer networks has become a hot topic of concern.As an emerging security technology,intrusion detection(Intrusion Detection)has gradually developed into a key technology to ensure the security of network systems.However,current intrusion detection systems generally have the problems of low detection performance and high false detection rate.At the same time,the network system structure is increasingly complex,the distributed environment is widely used,and the application of mass storage and high-bandwidth transmission technologies and the emergence of new attack methods continue to occur.In particular,some cooperating intrusions continue to emerge,bringing new issues to the research in the field of intrusion detection.This paper identifies network attack type detection and anomaly detection based on the data of network connection information,focuses on analyzing the advantages and disadvantages of network intrusion detection methods based on random forest algorithm,and proposes a random forest model based on optimized Drosophila and a random forest after clustering optimization.Model network data type detection methods to achieve the goal of improving the accuracy of network data type detection methods.The main tasks include:For the problem that the accuracy of data type detection of network connection information is not high,a random forest detection method based on fruit fly optimization is proposed.By analyzing the relationship between the random forest model detection strength and the base classifier,the fruit fly optimization algorithm is used to optimize the number of base classifiers and the number of selected splitting attributes to achieve the goal of improving the accuracy of network intrusion detection methods.For the problem that the accuracy of detection of random forest network intrusion type after fruit fly optimization is not high,a random forest detection method based on clustering optimization is further proposed.By analyzing the correlation of the base classifier of the random forest model and using the similarity clustering algorithm,the base classifiers that meet the metrics are aggregated together,the differences between the base classifiers are increased,and less base classifiers are used to achieve.Improve the goal of the accuracy of network intrusion detection methods.In order to make the network data type detection experiment results more objective and realistic,the experiment adopts the network transmission data in the real network environment,and uses the accuracy,mean square error,Confusion matrix,precision rate,recall rate and other evaluation parameters to synthesize the experimental results.Evaluation to achieve the purpose of improving the accuracy of network intrusion detection.