Network Intrusion Detection with Bat Algorithm for Synchronization of Feature Selection and Support Vector Machines
In order to improve the detection rate of network intrusion, this paper proposes a kind of bat algorithm (BA), which can optimize the intrusion detection model of support vector machine (BA-SVM). In this algorithm, parameters of the SVM support vector mac
- PDF / 224,483 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 57 Downloads / 188 Views
Abstract. In order to improve the detection rate of network intrusion, this paper proposes a kind of bat algorithm (BA), which can optimize the intrusion detection model of support vector machine (BA-SVM). In this algorithm, parameters of the SVM support vector machine are coded as individual bats first, and the detection rate of network intrusion is put as the parameter objective function. Then, the optimum parameter of support vector machine is found by simulating the bat flight. Finally, a network intrusion detection model is established based on optimal parameters, and simulation experiments are performed with KDD CUP99 dataset. The results show that this model could not only improve the detection rate of network intrusion, but also reduce the training time, and therefore improve the effect of network intrusion detection. Keywords: Bat algorithm vector machine
Network intrusion
Feature selection
Support
1 Introduction With the growing size and openness of the network, the number of cyber attacks and the degree of damage showed an upward trend. The traditional security measures can not satisfy the requirements of modern network security. As an active network security measure, network intrusion detection system has become a research focus in the field of network security [1]. Essentially, network intrusion detection is a multi-classification problem and mainly includes key procedures such as feature selection, classification parameters optimization etc. In practical applications, in order to solve SVM parameter optimization problems better, researchers have proposed genetic algorithms [2], PSO [3], AFSA [4] and other parameters to perform optimization of SVM parameters. In order to solve the conundrum of classification parameters and features set selection in network intrusion, this paper proposes a network intrusion detection method of synchronization of feature selection and SVM parameters based on bat algorithm (BA-SVM) and verify the effectiveness of BA-SVM with KDD CUP 99 dataset. © Springer International Publishing Switzerland 2016 L. Cheng et al. (Eds.): ISNN 2016, LNCS 9719, pp. 401–408, 2016. DOI: 10.1007/978-3-319-40663-3_46
402
C. Cheng et al.
2 Bat Algorithm Bat algorithm (BA) is a new type of stochastic optimization algorithm for the optimal solution of swarm intelligence search proposed by Yang from University of Cambridge in 2012 [5]. It is simple and easy to implement this algorithm, and it has become a hotspot of heuristics in recent years. Bat algorithm is a kind of optimization technique based on iteration. In the algorithm, all solutions are initialized to a group of random ones, and then the optimal solution is found through iterative search. And part of the new solution is generated around the optimal solution with randomly flight, which can enhance the local search. Compared with other algorithms, bat algorithm is superior to them in terms of accuracy and validity. Bats can explore for preys and avoid obstacles with its echolocation function, and they can find their habitats in t
Data Loading...
