Research And Application Of Network Intrusion Detection Technology Based On Active Learning Support Vector Machine

With the continuous development and maturity of Internet technology,network interaction plays an increasingly important role in people's daily life.How to model and analyze the historical network connection data,and on this basis,intrusion detection of the newly generated data has become a research hotspot.This paper designs and implements a multi-classification algorithm based on improved active learning support vector machine,and then designs and implements network intrusion detection application software based on this algorithm.It provides the functions of establishing classification model for representative samples,intrusion detection for network connection data detected and displaying detection results.In the process of selecting learning samples,the traditional active learning support vector machine classification algorithm only considers the distance between samples and classification boundary,but does not consider the redundancy between samples.To overcome this shortcoming,this paper proposes an improved active learning support vector machine binary classification algorithm by improving the learning strategy.The algorithm defines the sample selection degree,considers the distance between the sample and the current classification boundary,and the redundancy between the sample and the selected sample,which makes the selected sample more reasonable.In addition,by introducing the concept of sample selection width,the influence of sample selection number on the performance of the algorithm in each iteration process is analyzed.The experimental results show that,compared with the traditional active learning support vector machine classification algorithm,the proposed algorithm needs fewer labeled samples with the same prediction accuracy,and the convergence speed of the algorithm is faster.Traditional support vector machine(SVM)classification algorithms are usually used for data binary classification,which can not meet the need of diversity of attack types in network connection data.This paper extends the support vector machine binary classification algorithm and realizes the multi-classification algorithm based on support vector machine.Furthermore,on the basis of multi-classification algorithm of support vector machine,combined with active learning algorithm of improved learning strategy,the improved active learning multi-classification algorithm of support vector machine is realized.Through experimental analysis,in the case of multi-class samples,the improved active learning support vector machine multi-classification algorithm can use a small number of labeled samples to achieve high prediction accuracy.On the premise of designing and implementing an improved active learning support vector machine multi-classification algorithm,this paper designs and implements a network intrusion detection application software based on the project integrity.The application software can build an intrusion detection model by providing training samples,then use the detection model to detect the file data uploaded by users,and then display the test results in the form of tables,and provide the function of filtering the results by keywords.