Phishing Attacks Detection using Deep Learning Approach

Abstract

In the COVID-19 pandemic, people are enforced to adopt ‘work from home’ policy. The Internet has become an effective channel for social interactions nowadays. Peoples' immense dependence on digital platform opens doors for fraud. Phishing is a type of cybercrime to steal users' credentials from online platforms such as online banking, online business, e-commerce, online classroom, digital marketplaces, etc. Phishers develop fake webpages alike the original one and send spam emails to hook the users. Phishers seize users' credentials when an online user visits the counterfeit webpages through the spams. Researchers have introduced enormous tools like blacklist, white-list, and antivirus software to detect phishing webpages. Attackers always devise creative ways to exploit human and network weakness to penetrate cyber defense. This paper presents a data-driven framework for detecting phishing webpages using deep learning approach. More precisely, a multilayer perceptron, which is also referred as a feed-forward neural network is used to predict the phishing webpages. The dataset was collected from Kaggle and contains information of ten thousand webpages. It consists of ten attributes. The proposed model has achieved 95% training accuracy and 93% test accuracy.