Fuzzy integrated rough set theory situation feature extraction of network security

The element extraction from network security condition is the foundation security awareness. Its excellence directly disturbsentire security system performance. In this paper we introduce fuzzy logic based rough set theory for extracting security conditional factors. The traditional extraction method of network security situation elements relies on a lot of prior knowledge. With the purpose of solving this issue, in this paper we proposed fuzzy rough set theory based feature rank matrix of neighborhood rough set. Additionally, we propose reduction based parallel algorithm that uses the concept of conditional entropy in order to constructs the feature rank matrix as well as, constructs the core attribute by using reduction rules, takes the threshold of standard deviation as the threshold, and redefines the multi threshold neighborhood of mixed data. The attack type recognition training is carried out on lib SVM, filtered classifier, j48 and random tree classifiers respectively. The results demonstrate that the proposed reduction based parallel algorithm can increase the accuracy of classification, shorten the modeling time, and show increased recall rate and decreased false alarm rate.