Comparison of Performance of Phishing Web Sites with Different DeepLearning4J Models

Abstract

Due to the new type of coronavirus (COVID-19) disease, which was first seen in Wuhan, China in 2019, a pandemic was declared by the World Health Organization (WHO) on March 11, 2020. The pandemic, which is still in effect throughout the world, has changed the daily life activities and habits of the whole world community in a short time and shifted them towards digital media applications. Accordingly, increasing cyber-attack attacks and data breaches have created great risk for the pandemic society. In this context, the security of digital media applications has become a much more important issue with the COVID-19 outbreak. The issue has been observed especially on phishing websites. Web phishing is the practice of stealing personal information such as name, last name, password, and credit card numbers of individuals by imitating a reputable business. It will result in the exposure of information and the financial damage. The focus of the study is based on several DeepLearning4j (DL4j) models used to identify phishing websites. However, the main purpose of the study is to efficiently monitor the effectiveness of DeepLearning4J (DL4J) models with the aim of improving the performance of evaluation metrics.