Abstract
ABSTRACT With the recent advancement of the Internet of Things (IoT) in various sectors, security has become an essential requirement. Any IoT application or device may be compromised by intruders to disrupt the entire network. These kinds of insider attacks are difficult to prevent. Here, an Intrusion Detection System (IDS) can play an important role in identifying various unknown attacks. IDS uses network traffic logs to detect and respond to suspicious activities or anomalies before attackers exploit system weaknesses. Machine learning models are among the most efficient and effective methods to identify anomalous network behaviors. Hence, in this paper, we have conducted a comprehensive analysis utilizing several supervised and semi-supervised machine learning algorithms to assess their performance. We utilized 15 benchmark datasets containing network traffic samples related to various network attacks. We employed holdout and k-fold cross-validation methods for performance comparison. We also discussed the performance of the algorithms and identified possible reasons for their respective outcomes. Experimental results indicate that two supervised algorithms, kNN and ANN, exhibit the highest performance in terms of accuracy, precision, recall, etc. This comprehensive analysis of datasets and algorithms with various evaluation metrics provides researchers with valuable insights.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call