Combining TF-IDF, V-GAN, and XGB to Improve Next-Generation Web Application Firewalls' Ability to Detect SQL Injection Attacks

Combining TF-IDF, V-GAN, and XGB to Improve Next-Generation Web Application Firewalls' Ability to Detect SQL Injection Attacks

연구 분야: Databases

학회: 2024 IEEE/ACS 21st International Conference on Computer Systems and Applications (AICCSA)

SQL injection attacks are one of the most devastating vulnerabilities on the web, which can be used to leak sensitive information, gain unauthorized access, and result in financial losses. We utilized five distinct Machine Learning (ML) and Deep Learning (DL) algorithms to detect SQL injection attacks. We extracted features from network traffic and SQL queries through tokenization and regular expressions. Our findings show that combining the Term Frequency-Inverse Document Frequency (TF-IDF), Vanilla Generative Adversarial Network (VGAN), and eXtreme Gradient Boosting (XGB) models resulted in remarkably high performance metrics, including 99.95 % accuracy, 99.92 % precision, 99.97 % recall, and a 99.95 % F1-score. The results not only show the combination's effectiveness but also emphasize the importance of using machine learning techniques to detect SQL injection attacks to improve real-world cybersecurity applications significantly.