DeepFusionNet-NIDS: A Hybrid Deep Learning Fusion Model for Enhanced Network Intrusion Detection

As the exchange of sensitive data over digital networks grows, the need for sophisticated cybersecurity measures becomes increasingly critical. Network Intrusion Detection Systems (NIDS) serve as essential tools for identifying and mitigating malicious network activities. Traditional NIDS, typically reliant on signature-based detection methods, struggle to effectively detect novel and complex cyber threats. This research investigates the application of deep learning models to enhance NIDS accuracy and adaptability, with a specific focus on minimizing false positives. A comparative analysis of traditional machine learning techniques—including Random Forest (RF), Support Vector Machine (SVM), and Decision Tree (DT)—and deep learning models such as Convolutional Neural Networks (CNN), Artificial Neural Network (ANN), Deep Neural Networks (DNN), and Shallow Neural Network (SNN) is conducted using the KDD CUP 1999 dataset. For more experiment, we introduce a fusion model combining these deep learning architectures. Our proposed fusion model DeepFusionNet-NIDS achieves a test accuracy of 99.85%, outperforming individual models and illustrating the efficacy of hybrid approaches. These results suggest that deep learning-based NIDS offer a more robust solution for detecting both known and emerging threats, pointing to promising directions for future research.