Phishing Website Detection using Deep Learning
Md. Abu Ashraf Siddiq, Mohammad Arifuzzaman, Mazharul Islam
Abstract
Phishing attack is a type of cyber-attack where attacker sends fraudulent (spoofed, fake or deceptive) messages designed to lure a human victim to give away personal information or credentials or to deploy malicious software in victim's infrastructure like ransomware. As Internet usage is increasing day by day so are the cyber-crimes and scams. Phishing is one of the latest sophisticated techniques used by the scammers. Phishing website detection can help the users to avoid falling victim to these attacks. Although phishing websites are disguised as a legitimate one, fortunately they have some identifiable features. We have proposed a supervised learning approach using deep learning algorithms to detect phishing websites. We have achieved 94.8% accuracy using standard neural network model and achieved 93.6% accuracy with CNN (Conv2D) model. We have used a dataset downloaded from University of California, Irvine machine learning repository.