XGBoost Intrusion Detection Model Based On Improved WOA Optimization Parameters

At present has gotten in to the Internet era,which has become an absolutely part of real work and life there are malicious and aggressive behaviors in the network which make users' information security face huge challenges.As a means to accurately detect the attract date in the network intrusion detection has turned in to a plum in the field of information security.At present intrusion detection is the main means to detect attack dates from the perspective of improving the accuracy of intrusion detection this paper proposes a new intrusion detection model which is mainly based in extreme gradient boosting(XG-Boost)of machine learning and uses the optimization ability of whale optimization algorithm(WOA)to finds the best parameters In order to improve the optimization ability of conventional WOA algorithm an improved WOA optimization algorithm based on hybrid multi-strategy is proposed which is united with machine learning XGBoost and intrusion detection model based on WOA-XGBoost is proposed.The main research work is as follows:(1)Like other optimization algorithms the conventional WOA algorithm also has the problem of incomplete global search in the early stage and show convergence in the later local optimization which is easy to fall in to the problem of local optimization To solve this problem a variety of strategies are proposed to improved the conventional WOA algorithm.There are two main aspects of the original algorithm: the first is that the original algorithm is not completely improved in the early stage,and the local factor is not completely reduced in the later stage;Secondly,to solve the problem of poor balance between global optimization in the early stage and local optimization in the late stage,an adaptive weight strategy is introduced to change the step size of the algorithm.(2)In view of the problem that the traditional XGBoost parameter value method is poor and the algorithm performance is poor,the improved WOA algorithm is used to optimize the parameters of the XGBoost algorithm,an improved WOA-XGBoost algorithm is proposed,and an intrusion based on the improved WOA-XGBoost algorithm is established.Detection model.In terms of data processing,min-max normalization is used to denoise and unify the data,and then principal component analysis is used to process the initial network data and dimensionality reduction.The processed data is fed into the model to train the model.Finally,other data in the KDD CUP data set are randomly selected for testing.The results show that the improved WOA-XGBoost algorithm has significantly improved accuracy,sensitivity and specificity compared with the XGBoost algorithm with parameters set by the traditional method,and has a higher recognition rate.accuracy,which verifies the feasibility of the intrusion detection model.