Phishing URL Identification Using Machine Learning, Ensemble Learning and Deep Learning Techniques

Abstract

Demand for online applications and services is increasing day by day, whereas the security lapses in them is also becoming prominent. One of the major online threats, phishing, is consistently used to dupe users for retrieving sensitive information like bank account credentials, passwords, etc. Though so many techniques are getting introduced every day to detect phishing URLs, these attacks still persist. Hence, we conducted a survey to understand about the existing phishing detection techniques and their efficiency. Our study concluded that among the algorithms used in literature, random forest was achieving high accuracy. However, the most recent and extremely popular techniques like deep learning and ensemble learning were not explored sufficiently. In this paper, we propose to study the influence of deep learning and ensemble learning algorithms, along with well-known machine learning algorithms, namely naive bayes, random forest, decision tree and support vector machine on phishing URL detection. From our experimental results, we conclude that ensemble learning techniques are able to outperform other techniques.