Hybridized Machine Learning Prediction for the Exposure of Phishing Websites

Abstract

One of the most popular ways that people utilise the internet for illegal activities is phishing. Phishing websites are those that impersonate trustworthy websites while still appearing and sounding authentic. The purpose of their fabrication is to trick the receiver into believing that the item is authentic. These days, phishing schemes are riskier and more intricate than ever. Artificial intelligence-based ML & deep learning techniques can be used to anticipate phishing websites. A classification system based on ML may be employed to detect possible phishing websites. We provide a mixed machine learning approach for phishing site forecast in this study. The outcomes of the studies demonstrate that the recommended process performs more effectively in categorizing malevolent URLs than more contemporary methods. A graphical illustration of the accuracy comparison of several approaches is shown in Figure 4. The simulated results demonstrate that a compared to current method, the suggested hybrid classification strategy achieves higher accuracy. While the current KNN achieves 96% accuracy, the hybrid approach achieves 98.14% accuracy.