Year: 2023
Type: dissertation
Abstract: Cyber-attacks on computers and internet have been on a major increase since every organization has to communicate and run business operations on the network. The security awareness and protections implored by individuals and organizations daily is still not enough to fully eradicate the constant attacks. Bruteforce attack is a major high-level attack on the network connected via protocols like FTP, SSH or via the WEB. Brute Force attacks use methods like credential stuffing, DNS spoofing, multiple trials, and different consistent methods to access the targeted system or network. This attack type can be unpredictable and is often a gateway to many other attacks once successful hence, the need for a robust detection method arises. In this research, I investigated the application of Deep Belief Network (DBN) to detect Bruteforce attacks using the CSE-CICIDS2018 dataset. For additional performance comparison, we also trained three other algorithms: Decision Tree, Random Forest, and Logistic Regression. Experiment results showed that DBN model achieved a higher accuracy score of 99.3% while Random Forest had accuracy of 98.2%, Decision Tree 85.8% and Logistic Regression 99.2%.
Source: NORMA
Author: Diwe, Loveth Ukamaka
Institution: National College of Ireland
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7008988115
