Kwara State University
BUKOLA BALOGUN
Designation: LECTURER II
Department: Computer Science
| My Publications | ||||||||
|---|---|---|---|---|---|---|---|---|
| S/N | Title | Abstract | Authors | Volume Numbers | Publication Type | Publication Date | Link | |
| 1 | A Hybrid Metaheuristic Algorithm for Features Dimensionality Reduction in Network Intrusion Detection System. |
The advent of the Internet computer, and thus the amounts of connected computers in the last few decades, has opened vast quantities of intelligence to attackers and intruders. Firewalls are designed to identify, and block potentially harmful incoming traffic based on a predefined rule set. But, as attack tactics evolve, it becomes more difficult to differentiate anomalous traffic from regular traffic. Numerous detection strategies using machine-learning approaches have been suggested. However, there are issues with the high dimensional data of network traffic, the performance accuracy, and the high rate of false-positive and false-negative. In this paper, we propose a hybrid metaheuristic features dimensionality reduction method for Intrusion Detection Systems (IDSs). We used metaheuristic Bat algorithm for feature selection. The Bat algorithm selects sixteen (16) attributes. Subsequently, RNS was used to obtain the residues of the sixteen features selected. Then, the PCA was used to get the residues by extracting it. The experimental analysis was performed on NSLKDD dataset. The proposed Bat-RNS + PCA + RF achieved 98.95% accuracy, sensitivity of 99.40%, and F-score of 97.70%. The findings were also benchmarked with existing studies and our results were superior. | xBalogun B.F., Gbolagade K.A., Arowolo M.O., Saheed Y.K. | vol 12957 | Book Chapter | 2021-09-10 | https://link.springer.com/chapter/10.1007/978-3-030-87013-3_8 | |