The 2021 World Congress in Computer Science, Computer Engineering, and Applied Computing; CSCE 2021 Book of Abstracts, p. 106.
ISSN/ISBN: Not available at this time. DOI: Not available at this time.
Abstract: The CIC-IDS2017 and CSE-CIC-IDS2018 are state-of-the-art network intrusion data sets. They are typified by high dimensions and high volumes of benign network traffic and low volumes of network intrusion attacks, including sophisticated brute-force attacks. The imbalance and high-dimensional nature of such big data sets challenge machine learning algorithms used for network intrusion detection, particularly for feature selection. This paper adopts Benford’s Law as a feature selection method, since current research shows it can detect network intrusion attacks. The method was chosen based on observations about the frequency distribution of leading digits of network traffic data records. The paper investigates whether features identified by Benford’s Law significantly impact the performance of a popular semi-supervised anomaly detection Gaussian Mixture Model.
Bibtex:
@inproceedings{,
AUTHOR={Innocent Mbona and Jan H. P. Eloff},
TITLE={Evaluating a Semi-Supervised Intrusion Detection Algorithm Through Benford's Law},
BOOKTITLE={CSCE 2021 Book of Abstracts},
ADDRESS={Las Vegas},
MONTH={July},
YEAR={2021},
URL={https://www.american-cse.org/static/CSCE21%20book%20abstracts.pdf},
}
Reference Type: Conference Paper
Subject Area(s): Computer Science