Reference of Zeblok's AI-PaaS in
Proc. ACM Meas. Anal. Comput. Syst., Vol. 37, No. 4, Article 111. Publication date: August 2020
Applying Visual Adversarial Learning on VisualNet dataset for Robust Phishing Detection:
DYLAN CHOU, Carnegie Mellon University
AMIR RAHMATI, Stony Brook University | Ethos Lab Lead | Assistant Professor
The internet and computer networks brought about a platform where those with malicious intent can steal victim’s information through phishing. In turn, phishing detection techniques have been used to mitigate the number of successful phishing attacks. One means of detection involves the development of a machine learning model to predict when a suspicious webpage or email is indeed a phishing scheme and warn the victim. Throughout the literature, there has been a plethora of visual similarity-based, fuzzy data mining, and text mining approaches to phishing detection using machine learning, but a lack in adversarial approaches to bring out the weaknesses in recent phishing datasets. In this paper, a taxonomy of phishing webpage detection methods, including those that are adversarial, and input from adversarial sample generation will provide further robustness to the models fit to the datasets analyzed in the paper.