摘要
针对多类分类训练样本少、多类样本分布不均导致的入侵检测分类模型准确率低的问题,提出采用模糊支持向量机与多类逻辑回归相结合的2级入侵检测模型.模糊支持向量机(FSVM)1级检测模型将数据分为正常型和攻击型,多类逻辑回归(MLR)2级检测模型给出攻击型数据的具体类别.在模型设计中,给出了隶属度函数的计算方式,数据离散化、标准化和归一化的计算过程,以及MLR模型流程分析.实验证明,MLR模型比多种分类器分类准确率高,且耗时较短.FSVM-MLR 2级模型比MLR 1级模型准确率高.
There are two difficulties in intrusion detection multiclass classification model： fewer training samples and uneven distribution of training samples based on different classes. A two-stage multiclass classification model used in intrusion detection system is proposed, which combines Fuzzy Support Vector Machine and Multiclass Logistic Regression. FSVM is the first stage, in which the data is divided into two categories, normal and abnormal. MLR is the second stage, which can determine the type of abnormal data. The FSVM-MLR model provides the calculating method of fuzzy membership, the process of normalizing data set and the flow of MLR classification model. The result of simulation shows that, contrasting with some other classifiers, the MLR model has higher accuracy, and uses shorter training time. FSVMMLR two-stage model has higher accuracy than MLR.
作者
金志刚
苏菲
Jin Zhigang;Su Fei(School of Electrical and Information Engineering, Tianjin University, Tianjin 300072, Chin)
出处
《南开大学学报（自然科学版）》
CAS
CSCD
北大核心
2018年第3期1-6,共6页
Acta Scientiarum Naturalium Universitatis Nankaiensis
关键词
入侵检测
多类分类
模糊支持向量机
逻辑回归
intrusion detection
multiclass classification
fussy support vector machine
logistic regression
