Review of High Performance Network Intrusion Detection Engine

As the proliferation of connected devices and services increases, so does the demand for protective measures against cyber-attacks. Intrusion Detection Systems (IDS) are a crucial component of network perimeter security, detecting attacks by inspecting network traffic packets or operating system logs. While machine learning techniques have shown effectiveness in intrusion detection, few have utilized the time-series information of network traffic data, and none have included categorical information in neural network-based approaches. In this paper, we propose network intrusion detection models based on sequential information using Long Short-Term Memory (LSTM) networks and categorical information using embedding techniques. Our experiments on the UNSW-NB15 dataset demonstrate significant performance improvements, with binary classification accuracy reaching 99.72%.