Automated Prediction of Phishing Websites Using Deep Convolutional Neural Network

Abstract

Phishing is one of the ruinous issues encountered by the World Wide Web (WWW) and steers to the financial catastrophes for individuals and businesses. It has been perpetually a perplexing issue to identify phishing attacks with high exactness. The tremendous outcomes in the area of classification have been succeeded by the state-of-the-art invention of the deep convolutional neural networks (DCNNs). This paper is concerned with an accurate identifying approach for web phishing attacks based on deep convolutional neural networks. Our developed model has the ability to classify the attacked phishing websites from legitimate sites. However, due to the limitation of samples in the dataset, other machine learning algorithms (SVM, AdaBoost, Decision Tree, KNN) cannot perform proficiently for analyzing the data. In this respect, our proposed Deep Convolution Neural Network (DCNN) model has an automated approach to predict the phishing sites within the earlier stage. The empirical results show that the overall accuracy of 99% is achieved by the recommended methodology.