ABNORMAL INTRUSION DETECTION FOR EMBEDDED NETWORK SYSTEM BASED ON GENETIC ALGORITHM OPTIMISED SVM

Jiang Chunmao,Zhang Guoyin,Li Zhicong
DOI: https://doi.org/10.3969/j.issn.1000-386X.2011.02.087
2011-01-01
Abstract:As conventional intrusion detection method for embedded network system can't gain high detection accuracy,the network intrusion detection technology based on genetic algorithm optimised support vector machine(GA-SVM) is proposed in the paper.Support vector machine(SVM) classifier can well resolve non-linear classification problem with small training sample and high dimensions.However,the selection of training parameters of SVM has a great influence on classification accuracy.Genetic algorithm can consume less time to perform the optimisation of SVM parameters simultaneously.Thus,in the paper GA is used to synchronously optimise the SVM training parameters.Experimental results demonstrate that the GA-SVM classification intrusion detection model has very high detection accuracy.
What problem does this paper attempt to address?