The emergence of cybersecurity attacks as commonplace within the digital age calls for the development of cybersecurity defence capacities in both private and public organisations alike. Currently, machine learning models play a critical role in bolstering the defensive capacities of cybersecurity models. To determine the effectiveness of several machine learning models the NSL-KDD and TON_IoT datasets were utilised as benchmark cybersecurity datasets. These datasets allowed performance assessments in both binary and non-binary categorisations conditions. Model performance was judged based on multiple metrics such as accuracy, false-positive rate, and fit time. The results indicated that across both conditions that XGBoost outperformed several notable models such as Adaboost and random forest. Thus, based on the performance demonstrated with the NSL-KDD and TON_IoT datasets, XGBoost appears to be a promising cybersecurity machine learning method that could prove applicable to the classification of cybersecurity attacks.
History
Volume
1025
Pagination
93-111
ISSN
1860-949X
eISSN
1860-9503
ISBN-13
978-3-030-96629-4
Language
eng
Publication classification
B1 Book chapter
Editor/Contributor(s)
Ahmed M, Islam SR, Anwar A, Moustafa N, Pathan A-SK
Publisher
Springer
Place of publication
Berlin, Germany
Title of book
Explainable Artificial Intelligence for Cyber Security