基于模糊mars 的入侵检测系统 - 信息工程大学学报.pdfVIP

第 9 卷第 3 期 信息工程大学学报 Vo l. 9 No.3 2008 年 9 月 Journal of Information Engineering University Sep.2008 基于模糊 MARS 的入侵检测系统 程翔，江伴东，李毅成 (景德镇陶瓷学院信息工程学院，江西景德镇 333001) 摘要:多元自适应样条回归是一种有效的针对高维数据回归建模方法，将该算法应用于入侵检 测系统，根据入侵检测中存在的噪声数据和异常数据问题，提出了基于模糊算法的多元自适应 样条回归方法。通过基于 KDD1999 数据集的训练和测试，与 SVM 在数据集上的测试结果进 行对比，得出结论:该算法在入侵检测应用方面优于 SVM 算法。 关键词:入侵检测系统;多元自适应样条回归;支持向量机;模糊多元自适应样条回归 中图分类号:TP393.08 文献标识码:A 文章编号: 1671 - 0673 (2008 )03 - 0364 - 03 Intrusion Detection System Based on Fuzzy Multivariate Adaptive Regression Splines CHENG Xiang ,JIANG Ban-dong ,Ll Yi-cheng (Institute of Information Engineering , Jingdezhen Ceramic Institute ,Jingdezhen 333001 ,China) Abstract: Multivariate adaptive regression splines method is presented for flexible regression model- ing of high dimensional data , and multivariate adaptive regression splines method based on fuzzy a- rithmetic is present dealing with the noisy data and abnormal data in the intrusion detection. All the training and testing datasets are based on the KDD1999 intrusion evaluation data set. Multivariate a- daptive regression splines is trained and tested. The test results of SVM in data set are compared with that with multivariate adaptive regression splines. It draws conclusion that multivariate adaptive re- gression splines algorithm is better than SVM for intrusion detection. Key words: intrusion detection system ( IDS); multivariate adaptive regression splines ( MARS) ; suppo民 vector machines ( SVM) ; fuzzy multivariate adaptive regression splines ( FMARS) 入侵检测是一种通过收集和分析被保护系统 1 支持向量机简介 信息，从而发现入侵的技术。它的主要功能是对网 络和