APPLICATIONS OF SEQUENCE-TO-SEQUENCE AUTOENCODER NETWORKS IN REQUEST ANOMALY DETECTION
Keywords:
WAF, LSTM, seq2seq, autoencoderAbstract
This paper provides an insight into utilizing machine learning techniques to improve web application firewall (WAF) performance. A brief overview of existing techniques is provided, and a solution is proposed to optimize security breach alerts and anomaly detection capabilities of WAF software. An existing seq2seq autoencoder architecture is applied to solve the problem of efficient attack detection in WAF software.
References
Nginx WAF. https://docs.nginx.com/nginx-waf. Accessed 12 May 2019
Mod_security documentation. http://modsecurity.org/rules.html. Accessed 12 May 2019
Web Application Firewall (WAF) Evasion Techniques #2. https://medium. com/secjuice/web-application-firewall-waf-evasion-techniques-2-125995f3e7b0. Accessed 12 May 2019
A. Murzina, I. Stepanyuk, F. Sakharov, A. Reutov. Detecting Web Attacks with a Seq2Seq Autoencoder. https://habr.com/en/company/pt/blog/441030/. Accessed 12 May 2019
Open Web Application Security Project documentation. https://www.owasp.org/ index.php /Web_Application_Firewall. Accessed 12 May 2019
A Gentle Introduction to LSTM Autoencoders. https://machinelearningmastery. com/lstm-autoencoders/. Accessed 12 May 2019
