IntruVisor: A Novel CNN-Powered System for Network Intrusion Detection

Due to the complexity and large volumes, traditional intrusion detection systems find it very challenging to keep up high efficacies facing rapidly evolving cyber threats. This paper presents a novel Advanced Convolutional Neural Network-Based Intrusion Detection System: ACNN-IDS. This proposed system has been especially good at detecting complex attack patterns by automatically learning spatial and temporal features from data of network traffic using CNNs. ACNN-IDS architecture leverages multiple convolutional layers to extract complex representations of data. It then incorporates pooling and fully connected layers for classifying, with high accuracy, network events. It is trained by a diversified dataset that includes all types of network attacks to ensure the robust detection of the system under different scenarios. Experimental results have revealed that ACNN-IDS is capable of much better performance compared with traditional machine learning-based IDS in terms of detection accuracy, false positive rate, and computational efficiency. What makes this system very promising for the security of modern network infrastructures is that it is highly adaptive to new threats and can perform well on large datasets.