Abstract:
This project is focused on advancing the WAF (Web Application Firewall) technology on the basis of incorporating instant forecast using deep learning models – LSTM (Long Short-Term Memory) and RNN (Recurrent Neural Networks). Such models are highly useful for processing data streams in chronological order with engaging malicious web requests immediately before threats like DoS, Probe, R2L, U2R are realized.
We took advantage of the NSL-KDD dataset, a well-known benchmark in network intrusion detection systems, to train and test them. Proved itself to be a very strong pattern recognize during both the training and validation cycles, demonstrating its ability to learn and predict wisely. The project's effort to apply real-time prediction to the web-application firewall technology as a whole indicates a substantial move in the direction of strong and adaptable cybersecurity boundaries.
