download0 view131
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
Toward Bulk Synchronous Parallel-Based Machine Learning Techniques for Anomaly Detection in High-Speed Big Data Networks
Author(s)
Kamran Siddique김양우Zahid Akhtar김웅섭이행곤
Publication Year
2017-09-19
Abstract
Anomaly detection systems, also known as intrusion detection systems (IDSs), continuously monitor network traffic aiming to identify malicious actions. Extensive research has been conducted to build efficient IDSs emphasizing two essential characteristics. The first is concerned with finding optimal feature selection, while another deals with employing robust classification schemes. However, the advent of big data concepts in anomaly detection domain and the appearance of sophisticated network attacks in the modern era require some fundamental methodological revisions to develop IDSs. Therefore, we first identify two more significant characteristics in addition to the ones mentioned above. These refer to the need for employing specialized big data processing frameworks and utilizing appropriate datasets for validating system’s performance, which is largely overlooked in existing studies. Afterwards, we set out to develop an anomaly detection system that comprehensively follows these four identified characteristics, i.e., the proposed system (i) performs feature ranking and selection using information gain and automated branch-and-bound algorithms respectively; (ii) employs logistic regression and extreme gradient boosting techniques for classification; (iii) introduces bulk synchronous parallel processing to cater computational requirements of high-speed big data networks; and; (iv) uses the Infromation Security Centre of Excellence, of the University of Brunswick real-time contemporary dataset for performance evaluation. We present experimental results that verify the efficacy of the proposed system
Keyword
anomaly detection; network intrusion detection systems; bulk synchronous parallel; machine learning; big data; ISCX-UNB dataset; DARPA; KDD Cup 99
Journal Title
SYMMETRY-BASEL
Citation Volume
9
ISSN
2073-8994
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14720
http://www.ndsl.kr/ndsl/search/detail/article/articleSearchResultDetail.do?cn=NART89557087
Export
RIS (EndNote)
XLS (Excel)
XML

Browse