Deep-Learning Based Injection Attacks Detection Method for HTTP

In the context of new era high digitization and informatization, emergence internet artificial intelligence technologies has profoundly changed people’s lifestyles. The traditional cyber attack detection become increasingly weak in complex network environment era, deep learning technology begun to play a significant role field security. There are many kinds attacks against web applications, which very harmful, including SQL (Structured Query Language) injection, XSS (Cross-Site Scripting), command injection. Based on injection attacks, this paper combines also proposes multi-classification method for attacks. We extract features URL (Uniform Resource Locator) request body HTTP (Hyper Text Transfer Protocol) requests combine build model Firstly, aiming at problem imbalanced distribution training samples low accuracy attack, sample generation is proposed. experimental results show that proposed ensures higher rate lower false alarms. Secondly, we propose more expressive feature fusion model, effectively extracted by with discrete manually. work effective compared single model. improved about 1%.