Efficient Detection of Phishing Attacks with Hybrid Neural Networks
Xiaoqing Zhang, Dongge Shi, Hongpo Zhang, Wei Liu, Runzhi Li · 2018
Many machine learning techniques and social engineering methods have been adopted and devised to combat phishing threats. In this paper, a novel hybrid deep learning model is proposed to identify phishing attacks. It incorporates two components: an autoencoder (AE) and a convolutional neural network (CNN). The AE is adopted to reconstruct features that enhances correlation relationship among the features explicitly. The results from the experiments show that the model is able to detect phishing attacks with a mean accuracy over 97.68%, yet it has high generalization ability and can detect phishing attacks in the receivable time scale.