Research on Attributes Reduction Method of Intrusion Detection Data Based on Rough Set Theory. Issue 3 (October 2020)

• Record Type: Journal Article
• Authors: Fu, Xueliang; Zhang, Yaodan; Li, Honghui; Hu, Yumeng
• Is Part Of: Journal of physics. Volume 1624:Issue 3(2020)
• Publication Date: 2020-10
• Subjects: Big data -- Data mining -- Rough set -- NSL KDD -- Intrusion detection; Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1624/3/032036
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: With the huge amount of network connection data, high data dimensions and complex attack types, it is difficult for traditional data analysis techniques to obtain satisfactory results in the Network Security Intrusion Detection System. This paper proposed a RSDB method based on the Rough Set Theory and the Data Deblock Thought, which can help us to reduce the data conditional attributes and remix the data. The method considers two factors: the relationship between conditional attributes and decision attribute, and the data integrity. Through the experiment, this method can solve these difficult above problems effetely, while ensuring the original data is not distorted. Through many experiments and their experimental results, RSDB can not only make the classification accuracy over than 90%, but also its average detection time is in the millisecond level.