摘要
针对支持向量机分类方法在小样本、非线性情况下具有较好的泛化性能的特点,结合入侵检测系统实时性和适应性的要求,提出了一种应用动态支持向量机的入侵检测系统,来提高SVM模型的分类精度,并详细介绍了系统训练集以及分类模型动态更新的方法。最后对系统进行了仿真验证。实验仿真表明,该系统可有效的提高入侵检测的准确率,改善由于数据集更新造成的SVM分类精度下降的情况。
Support vector machine classification approach is considered has good generalization performance especially in small number and non-linear of training samples.Combining with the real-time performance and adaptability of IDS,a novel IDS based on dynamic support vector machine is established.The updating method of system training datasets and classification model is described in detail.Finally,simulation experiment for this system is performed,the result shows that this system can effectively improve the classification excursion caused by datasets updating,and enhance the true positive rate of the IDS.
出处
《计算机与数字工程》
2012年第11期118-120,共3页
Computer & Digital Engineering
关键词
支持向量机
入侵检测
数据集更新
support vector machine
intrusion detection
dataset updating