Application of Extreme Learning Machine (ELM) Classification in Detecting Phishing Sites

Abstract

Phishing site is a website created by internet criminals as closely as possible to resemble a real site to trick internet users by making it look like accessing a site from an official website. In overcoming the many phishing sites that exist in this study, the Extreme Learning Machine (ELM) classification method is used because ELM is one of the algorithms that is often used in classification and regression in machine learning. In this study, the accuracy value obtained from the test which was repeated 10 times was between 82-84% and the time between 5–11 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$s$</tex> with the best accuracy of 84.02% with a time of 7.98 <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$s$</tex> , the accuracy results generated from the ELM algorithm are indeed not very good. This large amount occurs because of the overfitting experienced by the formed classification model so that the false positives obtained are quite large. Referring to the dataset itself, the most influential feature or attribute in the labeling of phishing sites is the time domain expires, if the time domain expires has reached 200 days then the site has a phishing site label. In this study, ELM was compared with several other machine learning algorithms such as Support Vector Machine (SVM), Naive Bayes and Decision Tree.