Advanced Intrusion Detection Systems: Leverag...

Advanced Intrusion Detection Systems: Leveraging the Transformative Power of Deep Learning Algorithms for Enhanced Cybersecurity

Author Name : Manju J, Sumalatha M S

ABSTRACT Network intrusion detection (NID) is a significant method used in network systems to detect various cyberthreats. Traditional NID methods struggle with detecting unknown or new attacks, that necessitates a new approach. This study proposes a novel method based on Swin transformer; a deep learning (DL) architecture designed for enhancing intrusion detection systems (IDS) for cyber security applications. The Swin Transformer is ideally suited for intrusion detections because it effectively captures intricate patterns in network traffic. The NSL-KDD dataset, a widely recognized benchmark for NID that obtained from the Kaggle repository, is employed to test the model. The proposed approach uses the Swin Transformer to extract spatiotemporal features from the traffic data of dataset in order to categorize network activities into normal and attack categories. The robustness of the model is demonstrated with the performance evaluation metrics. The model achieved an excellent accuracy of 96.02%, including a precision of 96.01%, an F1-score of 95.99%, and a recall of 95.99%, highlighting its effective reduction of false positives and negatives. This study provides the basis for future research on utilizing deep learning techniques in cybersecurity applications while demonstrating the potential of transformer-based systems for intrusion detection.