Enhancing Phishing Website Detection Using Binary Multi-Objective Manta Ray Foraging Optimization: Binarization Rules Effects

Abstract

Phishing websites represent a significant cybersecurity threat by deceiving users into revealing sensitive information. This study introduces a novel multi-objective approach for feature selection aimed at enhancing phishing website detection. Unlike traditional single-objective optimization methods, the proposed approach simultaneously minimizes the error rate and minimizes the number of selected features, thereby addressing the trade-off between detection performance and computational efficiency.

In this paper, we propose an improved Binary Multi Objective Manta Ray Foraging Optimization (B-MOMRFO) algorithm to address the feature selection problem in phishing website detection. The primary focus is on analyzing the effectiveness of various Transfer Functions combined with different Binarization Rules in this context. The proposed approach is evaluated using two widely recognized phishing website datasets: UCI Phishing Websites Dataset(D1) and Mendeley dataset (D2).

Experimental results demonstrate that the proposed approach surpasses conventional classification methods by achieving higher accuracy with a reduced feature set. Furthermore, the findings underscore the importance of selecting appropriate Transfer Function and Binarization Rule combinations, as certain configurations yield superior performance in phishing website detection.