An Intrusion Detection Method Based on Support Vector Machine Hybrid Feature Selection

Abstract

The invention relates to the field of intrusion detection network security, in particular to an intrusion detection method based on support vector machine hybrid feature selection. The optimal feature subset is used to complete real-time intrusion detection, including: importing intrusion detection data sets; Initialize the parameters to get the optimal features; use the improved GSA algorithm to optimize the penalty parameters and kernel function of the Gaussian kernel SVM to obtain the SVM classifier; use the recursive features of the optimized SVM to sort the optimal features; use the improved gravity search The book algorithm obtains the initial population in the excellent subset, and obtains the optimal feature subset in the specified feature space according to the initial population; the optimal feature subset is input into the intrusion detection algorithm, and the improved SVM is used as the classifier for classification. Intrusion detection; the present invention improves the traditional gravitational search algorithm, namely, introduces the Tent chaotic sequence, which avoids the problem of premature convergence of the gravitational search algorithm.

Description

technical field [0001] The invention relates to the field of intrusion detection network security, in particular to an intrusion detection method based on support vector machine hybrid feature selection. Background technique [0002] In view of the security threats faced by the network, ensuring network security has become a prerequisite for the healthy and stable development of network informatization in various fields. The current network protection measures are mainly divided into two aspects: technology and management. In terms of technology, identity authentication and access control technology, network firewall technology, file encryption technology and intrusion detection technology are mainly used. As the existing protective measures are mostly preventive. Even if the existing defense technology can better protect network security, knowing the specific types of intrusion data is still an important reference for designing a reasonable network protection model. [00...

AI Technical Summary

Problems solved by technology

[0004] However, the network intrusion data to be detected is often high-dimensional data. When detecting a large amount of high-dimensional intrusion data, this method has the disadvantages of low detection speed and high false alarm rate.

Images