点击次数：

论文类型：期刊论文

发表时间：2012-02-01

发表刊物：Journal of Convergence Information Technology

收录刊物：EI、Scopus

卷号：7

期号：3

页面范围：84-90

ISSN号：19759320

摘要：Network attack detection is an important aspect in maintaining a stable network. It is usually done by a rule-based method, such as firewall technology. In the case of network data stream, rule-based method cannot be changed with the help of the outside data dynamic environment. In this paper, we proposed an incremental learning method based on support vector machine (SVM), which can improve the classification accuracy as well as reduce the time-consuming aspect. The validation of our method compares with two other models: k-nearest neighbor algorithm (KNN) and decision tree. The experimental results show the robustness of the proposed model.