Abstract
The problem of identifying threats in cyberspace is basically a classification problem for machine learning, which is also a need of the hour. While a large number of works related to vulnerability detection have been reported recently, the majority of the works have been focused on IT systems. Various factors related to these systems are considered for the classification of risk levels in these works, and appropriate recommendations about expected risks and vulnerabilities in the future will be given to their users. This work aims to detect possible risks and vulnerabilities and notify the user in a timely manner using parameters related to the user’s working PC in addition to the system on which the user relies. Various system-level and hardware-level features are extracted. The extracted features are clustered into three target classes: low, medium, and high. The obtained cluster groups are then fit into the regression model designed using linear regression, LASSO regression, and support vector regression for training. On evaluation, all three methods achieved almost equally in prediction, with support vector regression taking a slight lead in predicting the potential attack on the user.
Recommended Citation
Alshehri, Hamdan
(2025)
"Ensemble Learning-Based Vulnerability Detection and Forecasting,"
University of Bisha Journal for Basic and Applied Sciences: Vol. 1:
Iss.
1, Article 6.
Available at:
https://ubjbas.ub.edu.sa/home/vol1/iss1/6