Loading...
Malicious URL and Intrusion Detection using Machine Learning
Hamza, Amr Hammam, Farah Abouzeid, Medhat Ahmed, Mohammad Arsalan Dhou, Salam Aloul, Fadi
Hamza, Amr
Hammam, Farah
Abouzeid, Medhat
Ahmed, Mohammad Arsalan
Dhou, Salam
Aloul, Fadi
Date
2024
Advisor
Type
Article
Peer-Reviewed
Preprint
Peer-Reviewed
Preprint
Degree
Citations
Altmetric:
Files
Loading...
ICOIN_paper_revised.pdf
Adobe PDF, 866.22 KB
Description
Abstract
Cyberattacks are becoming increasingly sophisticated and evolving danger to the Web users. Therefore, addressing the growing threat of cyberattacks and providing automated solutions became a necessity. The purpose of this paper is to use machine learning (ML) techniques for malicious websites detection and classification, and intrusion detection. Different ML algorithms were applied, namely Decision Tree (DT), K-Nearest Neighbors (KNN), Naive Bayes (NB) and Support Vector Machine (SVM). Two datasets were utilized to train the MLmodels. The first dataset contains two classes of websites: “malicious” and “benign”. The second dataset has six classes of different network intrusion cyber-attacks: “normal”, “blackhole”, “TCP-SYN”, “PortScan”, “Diversion”, and “Overflow”. Experimental results demonstrated that the ML algorithms were able to achieve high accuracy in predicting website maliciousness and intrusion detection. Using the first dataset, DT KNN, and SVM classifiers exhibited the best performance for detecting malicious URLs with accuracies over 99%. Using the second dataset, the DT classifier proved most suitable for intrusion detection, achieving an accuracy of 95%. This paper suggests the integration of ML techniques into online security systems to enhance their efficacy in detecting and preventing cyber threats.