Leveraging CNNs, Quantization, and Random Forest for Edge Deployable Intrusion Detection Efficiency

This research introduces a novel Intrusion Detection System (IDS) for edge computing, blending Convolutional Neural Networks (CNN) and Random Forests for feature reduction, with a focus on anomaly detection. The use of quantization techniques on ensembled CNN models is a key aspect, ensuring the system remains lightweight while preserving detection accuracy. This approach is particularly effective in resource-limited settings, responding to the need for efficient cybersecurity in such environments. The ensemble model, after quantization, achieved an accuracy of 98.13% and attained a 50% reduction in size, maintaining robust detection capabilities even in constrained resource scenarios. The study's results indicate a remarkable achievement in balancing computational efficiency with high detection performance, marking a significant step in edge-deployable cybersecurity technologies.