EXTREME LEARNING MACHINE FOR CLASSIFICATION OF PHISHING WEBSITES FEATURES

Phishing sites are websites designed by online criminals to closely mimic legitimate websites in order to deceive internet users into thinking they are viewing a legitimate website. This study makes use of the “Extreme Learning Machine” (ELM) classification technique in order to circumvent the multiple phishing sites. ELM is one of the algorithms that is frequently utilised in regression and classification techniques in machine learning. In this particular investigation, the accuracy results that were generated by the ELM algorithm are not, in point of fact, very good at all. The accuracy value that was obtained from the test that was carried out ten times ranged between 89 and 93 percent, and the time that was required ranged between 7 and 13 seconds, with the highest rate of accuracy being 90.72 percent and a duration of 8.77 seconds. . The overfitting of the classification model, which leads to the identification of a sizeable fraction of false positives, is responsible for the substantial number that was detected. Regarding the dataset itself, the time domain expires is the most important characteristic or attribute in the labelling of phishing sites; Phishing site status is assigned to a website when the domain's expiration date is more than 250 days in the future. The K-Nearest Neighbour and Random Forest machine learning methods were compared to ELM in this study