Research of Network Intrusion Detection Based on DF-FSVM

Abstract

Aiming at the low detection rate in network intrusion detection system, the paper presents a method of imbalanced data classification based on Double Factor FSVM (DF-FSVM). Considering the problem of imbalance and noise and isolated points in the training sample, the FCM clustering method is used to calculate the intra-class imbalanced factor to form the fuzzy membership function. The sample imbalance is caused by factors such as the number and the dispersion of samples. Therefore, inter-class imbalanced factors were introduced in the fuzzy membership function. And machine learning and classification were designed for imbalanced samples based on DF-FSVM. The experimental results show that this method can effectively improve the detection accuracy of intrusion detection system compared with standard support vector machine (SVM) and fuzzy support vector machine (FSVM).