Intrusion detection based on SVM and decision fusion
Rui-xia Zhang, Zhenrong Deng, Wenhui Zhang, Zhi Guo-jian · 2010
Feature selection and classifier are two important issues in intrusion detection to achieve high performance. This paper proposes intrusion detection scheme based on feature selection with different feature selection methods. Then the extracted features are employed by Support Vector Machine (SVM) for classification. But in fact, single classifier doesn't attain satisfying performance. To address the problem, independent classification outcomes are aggregated through different decision fusion strategy. To examine the feasibility of the scheme, several experiments have been done on dataset in KDD-99. Results indicate the high detection accuracy for intrusion attacks and low false alarm rate of the reliable system.