Phishing Email Detection Model using Improved Recurrent Convolutional Neural Networks and Multilevel Vectors

Abstract

The Phishing Email attacks are the most critical and commonly occurring type of cyber threats in thetoday’s world and it causes huge
financial losses.To reduce the effect of these attacks, we need a new phishing detection technology to control the phishing email
attacks. In this paper, we proposed a new detection technique named Improved Themis model. This model analyzes the email
structure based on email header and body.Our model includes Common Bag of Words (CBOW) as Multi-level Word2Vec
mechanism and Improvised Recurrent Convolutional Neural Networks(RCNN) algorithm withAttention approach. The experimental
results show the overall accuracy of the proposedmodel yields 99.87% and the negligible False Positive Rate (FPR) is 0.042%.The
promising results are effective than the existing identification techniques and verifies the model’s performanceto detect the phishing
emails.