Enhancing cybersecurity: A review and comparative analysis of convolutional neural network approaches for detecting URL-based phishing attacks

Abstract

Phishing attempts to mimic the official websites of businesses, including banks, e-commerce, government offices, and financial institutions. Phishing websites aim to collect and retrieve sensitive data from users, including passwords, credit card numbers, email addresses, personal information, and so on. The growing frequency of phishing attacks has prompted the development of numerous anti-phishing technologies. Because machine learning (ML) techniques perform better in categorization problems, they are used extensively. But the most crucial features are not extracted by the algorithms in use today, which could result in a false categorization. In addition, the complex algorithms contribute to the long reaction time. To solve these issues, this study suggests using a Bidirectional Long Short-Term Memory-based Gated Highway Attention Block Convolutional Neural Network (BiLSTM-GHA-CNN) to detect phishing URLs.

Keywords

• URL
• Phishing
• Features
• Neural network
• Machine learning
• Deep learning