Phishing Attack Detection Method Based on Deep Learning and Metaheuristic Pelican Optimization Algorithms

Abstract

In the cyberattacks, the phishing attack is the most harmful way used by attackers to steal the sensitive information of the individual or organizations. This attack is often accomplished through clicking or opening attachments through emails, website URLs, and social media. In the previous studies, the traditional approaches, namely, blacklisting, website content, and URL features, were employed for phishing attack detection. However, these approaches are slow and fail to detect the new phishing attacks. Thus, the machine learning (ML) algorithms are employed to overcome the previous issues. In this paper, we have designed a phishing attack detection method based on ML that efficiently detects the phishing attack by analyzing the website URLs data. Initially, the optimal feature selection from the data is done by combining the metaheuristic pelican optimization (PO) algorithm with artificial neural networks (ANN). After that, a lightweight Convolutional Neural Network (CNN) is employed to detect the legitimate and malicious website URLs. Furthermore, by dividing the standard dataset into various training and testing ratios, the suggested approach can be tested on those data. The result indicates that the proposed method effectively detects the phishing attack and outperforms the previous methods by achieving an average accuracy value of 0.9995.