Design And Implementation Of Integrated Intrusion Classifier Based On Machine Learning

With the rapid development of computer technology,intellectualization has become a trend.Although it can bring great convenience to people's daily life,intellectualization also means programmability,which may cause adverse consequences and immeasurable economic losses.Therefore,how to design an intrusion detection system is particularly important.This paper aims to improve the overall efficiency of intrusion detection and classify the types of intrusion.In view of the high dimensionality of data points in the original data set,this paper studies the related algorithms of machine learning,and designs a classification model of Intrusion Detection Based on machine learning.The key of intrusion detection is classification algorithm.At present,the research on machine learning is becoming more and more mature,and in Intrusion detection-related models,such as support vector machines.These classification models usually need a lot of high-quality data for model training to achieve better results.However,in the actual production environment,there are many problems,such as uneven sample distribution and too long model detection time.This paper is based on Support Vector Machine(SVM)algorithm,and studies the optimization of data processing and classification algorithm in order to obtain better detection results.In the network intrusion detection model proposed in this paper,the improved K-means algorithm is first introduced into the training data set selection.In the subsequent training process of SVM classifier,an automatic parameter optimization method based on grid search and simulated annealing is proposed.Finally,an intrusion detection model based on support vector machine(SVM)is constructed.The KDD99 data set is used to test the corresponding model.The proposed intrusion detection model can achieve better performance and effectively improve the overall detection rate.Compared with other intrusion detection methods,the model has better performance in detecting U2 R attacks and R2 L attacks with serious network hazards.At the same time,the model can also reduce the time of model training and testing,and has a certain generalization ability.