Scholarlink

Web applications are still at risk from SQL injection (SQLi), which calls for strong detection systems that can recognize nefarious SQL queries. This study investigates the use of two recurrent neural network (RNN) variants-Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models-for SQL injection attack detection. In contrast to conventional techniques, LSTM and GRU models are good at identifying long-term patterns and sequential dependencies, which makes them good options for improving SQL injection detection’s efficacy and accuracy. A comprehensive dataset with a variety of legal queries and SQL injection attack cases is used to assess the performance of the LSTM and GRU designs. The results show high levels of sensitivity, specificity, and accuracy, successfully reducing false positives while correctly recognizing possible threats.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SQL Injection Detection Using Recurrent Neural Networks (RNN)

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • V. Valli Kumari,
  • Y. Prasanna Kumar

摘要

Web applications are still at risk from SQL injection (SQLi), which calls for strong detection systems that can recognize nefarious SQL queries. This study investigates the use of two recurrent neural network (RNN) variants-Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models-for SQL injection attack detection. In contrast to conventional techniques, LSTM and GRU models are good at identifying long-term patterns and sequential dependencies, which makes them good options for improving SQL injection detection’s efficacy and accuracy. A comprehensive dataset with a variety of legal queries and SQL injection attack cases is used to assess the performance of the LSTM and GRU designs. The results show high levels of sensitivity, specificity, and accuracy, successfully reducing false positives while correctly recognizing possible threats.