Design and Analysis of Intrusion Detection Using Deep Learning Models
Abhishek Modak, Vasudev Dehalwar · 2024
This paper presents a comprehensive design and analysis of intrusion detection using two prominent intrusion datasets: NSL-KDD and CSE-CIC-IDS-2018. The deep learning, Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) models were used to detect intrusion detection. A hybrid CNN-GRU model that combines CNN and GRU architectures is proposed for intrusion detection. The performance of these models is rigorously analyzed by utilizing key metrics such as accuracy, precision, recall, f1-score, and Receiver Operating Characteristic (ROC) curves. Our experiment demonstrates an enhancement of 30% in intrusion detection with the CSE-CIC-IDS-2018 dataset compared to NSL-KDD. Our proposed hybrid CNN-GRU model outperforms individual CNN and GRU models by around 10% across all assessment measures. These results underline the importance of deep learning approaches for intrusion detection in evolving cyber threat scenarios.