Improving Phishing Detection with the Grey Wolf Optimizer

Abstract

With the recent epidemic of COVID-19-themed scam and phishing, the efficient automated detection of such attacks is crucial. Although many anti-phishing solutions, such as lists and similarity and heuristic-based approaches detect attacks, methods still can be improved. Classification accuracy is highly dependent on the feature selection method used to select appropriate features for classification. In this article, a multi-objective grey wolf optimizer is used to select proper features for classifying phishing websites through a variational autoencoder. Our results indicate the superiority of the classification rate compared with related work: A classification rate of 97.49%, is obtained, thereby suggesting the feasibility of evaluating our work.