Anticipating Malicious Server Attacks: Evaluating the Effectiveness of Various Machine Learning Models

Qenawy, Al-Seyday. T.; Ahsan, Muhammad

doi:10.21608/jaiep.2024.314522.1007

Anticipating Malicious Server Attacks: Evaluating the Effectiveness of Various Machine Learning Models

Document Type : Original Article

Authors

¹ Intelligent Systems and Machine Learning Lab, Shenzhen 518000, China

² School of Mathematical Sciences, Jiangsu University, Jiangsu 212013, China

10.21608/jaiep.2024.314522.1007

Abstract

The global shift to online payments means that companies face growing cyber dangers, especially to servers. The target of this analysis is on malicious server hacks to be forecasted based on anonymized incident data of several features that are logging parameters and an outcome variable of hack occurrence. Based on the problems context, several machine learning models were created and tested such as K-Nearest Neighbors, Naïve Bayes, Neural Networks, Gradient Boosting, and finally the e SVM with the RBF Kernel for the prediction of possible server hacks. The models were evaluated according to the performance indicators such as accuracy, sensitivity, specificity, precision, F1 measure. As for the models, the highest accuracy was recorded for K-Nearest Neighbors with 93.5\% while still revealing the highest sensitivity which makes it the best model in making a prognosis on server hacks. The second model, Neural Network, also demonstrated good results in terms of Sensitivity and F1-score. Based on our study, it is evident that these machine learning models can be used to predict possible future server hacks thus acting as a preventive measure in cybersecurity. This paper has explored the practical application of machine learning in cybersecurity and other related topics while the future work is expected to look at other advanced models and other features that would improve the recognition’s accuracy.

Keywords