Research And Implementation Of Intrusion Detection Technology Based On Feature Selection And Ensemble Learning

With the rapid development of computer network,more and more network intrusion events occur around us.From national security to user privacy,the impact of network intrusion attacks is becoming more and more extensive.At the same time,the ways of intrusion are constantly updated.The traditional firewall and encryption authentication have been difficult to prevent most intrusion attacks.In order to enhance the protection of network security,intrusion detection technology has been proposed and paid attention to.With the rise of artificial intelligence,more and more machine learning is applied in various fields,including intrusion detection technology.This paper focuses on the research and implementation of intrusion detection technology based on feature selection and ensemble learning.Through the combination of improved particle swarm feature selection method and deep stacking ensemble learning,the problems of low detection rate,poor effect and inability to detect new attacks of intrusion detection technology are solved.The main innovations of this paper are as follows:(1)This paper analyzes the current situation of intrusion detection,finds the causes of the problems by studying the public data set of NSL-KDD intrusion detection,and improves the intrusion detection technology from the data level.Feature selection is proposed to improve the detection success rate of machine learning in intrusion detection.In this paper,particle swarm optimization algorithm is used for feature selection,and an improved particle swarm optimization algorithm is proposed for feature selection.In the improved algorithm,the learning strategy of improving local particles is proposed,and the random search performance of particles is increased by adding the adaptive function PC.The improved algorithm improves the search performance,reduces the possibility of falling into local optimization,and can be better suitable for feature selection.We apply the improved feature selection to intrusion detection,which improves the detection accuracy and reduces the detection time.(2)Aiming at the problem that a single classifier has poor neutral performance in intrusion detection and the attack detection rate of some categories is too low,a deep stacking network is proposed as a learning strategy and model of ensemble learning.In the experimental part,the different detection characteristics of classifiers are analyzed,and the effects of each classifier are combined to complete the intrusion detection model based on ensemble learning.Using ensemble learning to improve intrusion detection classification can improve the detection performance and make up for the low detection rate of a single classifier in some types of attack detection.(3)An improved intrusion detection scheme is proposed,which uses two machine learning technologies proposed in this paper:feature selection and ensemble learning.The technology is applied to the intrusion detection scheme to improve the efficiency of intrusion detection.At the same time,a new intrusion detection framework is proposed,which combines misuse detection and anomaly detection.By learning the new attack type and adding it to the rule base,the new design framework can detect new attacks to a certain extent.The proposal of this scheme provides a strong support for the development of network intrusion detection technology.