DETECTION OF WEB PHISHING ATTACKS USING DEEP BELIEF NETWORK ALGORITHM

Abstract
The exponential increase and widespread usage of computers and internet enabled services, as well as growth in online financial activities in Nigeria has birthed huge concern in terms of proliferation of cyber crimes. Phishing attacks are one of the highest rated attacks launched on internet users in Nigeria. The introduction of Bank Verification Numbers in Nigeria however, brought a sporadic increase in rates of web phishing attacks. The paper proposes a web phishing detection system based on deep belief network algorithm. The aim of proposing this algorithm is due to the dynamism of the phishing attack across the cyber landscape this dynamisms initiate limitations in several systems studied and due to the peculiarity of the Nigerian attack landscape the researchers hence proposed thorough analysis of mail contents using Artificial Neural Networks (Deep belief Network Algorithm) to handle classification of spam words in addition to increasing the spam keywords (by including specific spam keywords used to scam Nigerian online users) in the database in other to achieve more accurate results. The proposed model will handle feature selection using the wrapped-based feature selection approach. Automation of feature selection of the model would cater for the dynamic strategic changes in phishing attacks

Keywords: Phishing, Machine Learning, Deep Belief Network, BVN.