Utilizing Logistic Regression to Detect Tautology-Based SQL Injection Attacks

Utilizing Logistic Regression to Detect Tautology-Based SQL Injection Attacks

연구 분야: Databases

학회: World Congress in Computer Science, Computer Engineering & Applied Computing

Cybersecurity threats increasingly jeopardize the digital landscape, necessitating advanced detection methods. This paper presents an AI-based Logistic Regression algorithm designed to identify Tautology-Based SQL Injection attacks targeting username and password inputs. Developed and tested using ML.NET, the algorithm achieved an 87% accuracy rate, outperforming the 80% accuracy of the Random Forest algorithm. Logistic Regression demonstrated notable effectiveness in this binary classification task, underscoring its capability to accurately distinguish between legitimate and malicious queries. The model’s high accuracy and robust performance in handling yes-or-no outcomes highlight its significant value in high-stakes cybersecurity environments. The optimization provided by the ML.NET framework further ensures the reliable detection of SQL injection threats.