Novel Approach to Phishing Detection Using ML and Visual Similarity

Abstract

Sanghavi, Preet Kunchapu, Achyuth Kulkarni, Apeksha Solani, Devansh Anson, A.Phishing involves attempts to trick the user by extracting crucial important information that should otherwise not be leaked. Data like bank account numbers, social media accounts, company revenue reports, online transactions secrets are some of the examples that an attacker hopes to extract by exploiting the vulnerability of the user. In today’s day and age, it is very important to warn the populace against such attacks and provide relevant security awareness. Through our research, we aim to provide a brief study and novel insights into how machine learning and image visualization can be used to detect phishing webpages using a two-part implementation technique. The first part involves an analysis and overview of different machine learning classifiers across different datasets and features to find the most relevant technique to detect phishing websites efficiently. The second part introduces a novel approach of how discrete Fourier transform and image comparison can be used to compare suspicious webpages with real ones. This study aims at providing a new direction for researchers to understand how machine learning and automation can be inculcated in phishing detection to avoid any data breach.