Department of Computer Engineering, Kerman branch, Islamic Azad University, Kerman, Iran
10.24200/sci.2026.66452.10067
Abstract
The challenge of detecting and preventing phishing attacks is becoming increasingly difficult due to their growing sophistication. Phishing is a social engineering tactic where users are tricked into revealing sensitive information, such as passwords, credit card details, and usernames, through fake websites that appear legitimate. Traditional detection techniques often fail to keep up with the dynamic and evolving nature of phishing tactics, necessitating more adaptive and complex approaches. This study introduces a novel method for identifying phishing websites through a multi-agent deep learning approach. The method uses three deep learning networks, each trained to analyze specific features of websites, including the URL, page content, and the Document Object Model structure. A highest confidence score mechanism is then applied to combine the results of these networks and produce a final prediction. The results demonstrate that this approach outperforms current advanced phishing detection methods, achieving an accuracy rate of 99.20% and a false positive rate of just 0.20%. This proposed method not only excels at identifying known phishing sites but also shows strong performance in detecting previously unseen sites, making it highly adaptable to emerging phishing techniques. The approach has significant potential to improve web security by assisting both users and organizations in detecting and preventing phishing attacks. Furthermore, this framework highlights the potential of deep reinforcement learning in cybersecurity, paving the way for more resilient and automated security systems to combat the growing threat of cyberattacks.
Jafari, Z. , Ghafouri, S. H. and Ahmadinia, M. (2026). Adaptive Phishing Detection on Webpages via Multi-Agent Deep Learning and Multi-Dimensional Features. Scientia Iranica, (), -. doi: 10.24200/sci.2026.66452.10067
MLA
Jafari, Z. , , Ghafouri, S. H. , and Ahmadinia, M. . "Adaptive Phishing Detection on Webpages via Multi-Agent Deep Learning and Multi-Dimensional Features", Scientia Iranica, , , 2026, -. doi: 10.24200/sci.2026.66452.10067
HARVARD
Jafari, Z., Ghafouri, S. H., Ahmadinia, M. (2026). 'Adaptive Phishing Detection on Webpages via Multi-Agent Deep Learning and Multi-Dimensional Features', Scientia Iranica, (), pp. -. doi: 10.24200/sci.2026.66452.10067
CHICAGO
Z. Jafari , S. H. Ghafouri and M. Ahmadinia, "Adaptive Phishing Detection on Webpages via Multi-Agent Deep Learning and Multi-Dimensional Features," Scientia Iranica, (2026): -, doi: 10.24200/sci.2026.66452.10067
VANCOUVER
Jafari, Z., Ghafouri, S. H., Ahmadinia, M. Adaptive Phishing Detection on Webpages via Multi-Agent Deep Learning and Multi-Dimensional Features. Scientia Iranica, 2026; (): -. doi: 10.24200/sci.2026.66452.10067
)