Research On Intrusion Detection Method Based On Population Clustering And Feature Selection

The significant growth of data in the Internet era has provided great convenience for people’s various network needs on the one hand,and on the other hand,the current network is facing many security risks.The ever-expanding popularity of web applications has resulted in an exponential increase in the amount of data in the web.At the same time,the dimension of data has also increased rapidly,which has brought greater challenges to the research in the field of network intrusion detection.In order to effectively deal with the dimensional disaster and make better use of the data in the network,through the research and analysis of the existing feature selection methods and intrusion detection related technologies,an intrusion detection model based on population clustering and feature selection is proposed.The main work is as follows:(1)In order to solve the problem that the feature selection method is prone to premature convergence in the early search process,a clustering method based on population is proposed.First of all,the similarity function is defined from the Hamming distance and the accuracy difference between the particle and the cluster center,so as to calculate the similarity between the particle and the cluster center.Further more,the goodness is defined from the two aspects of feature location and classification accuracy.The initial particles are clustered by calculating the goodness value of the particle features in the cluster and using a threshold comparison method.The obtained surrogate particles are used as the initial population of the particle swarm optimization algorithm to perform a global search,so that the distribution of the initial population in the entire feature space is wider,and the final feature subset contains more promising features.The experimental results show that the method can effectively avoid the premature convergence problem in the early search and improve the performance of the classification model.(2)In order to solve the problems of low accuracy and detection rate in the field of intrusion detection,an intrusion detection model based on species clustering and feature selection is proposed.Firstly,the integrated method which combining the two feature selection methods of packing and filtering is adopted,and the mutual information method is integrated into the improved particle swarm optimization algorithm.In order to find the feature with the least redundancy with the selected feature and the greatest correlation with the target class,we calculate the mutual information between features and features and between features and classes,so as to realize the pruning of the feature subset obtained by the particle swarm algorithm.After that,a dual objective function including the maximum detection rate and the minimum false alarm rate is defined as the fitness function based on the above feature selection method.Then the intrusion detection model is constructed,so as to be more targeted.Perform intrusion detection.The experimental results show that the model has better performance in intrusion detection,effectively improves the detection accuracy and detection rate,and reduces the false positive rate of detection to a certain extent.