Sharma, Priynka and Sharma, Sushita (2026) Deep Learning Data - Driven Anomaly Detection for Zero - Day Malware Prevention in Cybersecurity. In: Proceedings of the Third Congress on Control, Robotics, and Mechatronics. Smart Innovations, Systems and Technologies, 448 . Springer Nature, Singapore, pp. 347-360. ISBN 978-981-96-8125-9
Full text not available from this repository. (Request a copy)Abstract
The detection and prevention of zero-day attacks represent a critical challenge in modern cybersecurity, particularly given the limitations of traditional signature-based methods that struggle to identify previously unseen threats. This research explores applying deep learning techniques to advance anomaly detection for zero-day malware detection. Specifically, the study uses Recurrent Neural Networks (RNNs) and autoencoders to model normal system behavior and detect subtle deviations that may indicate malicious activity. These deep learning models are designed to learn from vast datasets containing benign and anomalous system behaviors, ensuring their robustness and adaptability in diverse environments. A significant contribution of this work is developing a scalable framework capable of processing large volumes of data in real time, providing continuous monitoring for zero-day attacks. By integrating advanced feature engineering techniques, the proposed system enhances detection accuracy while minimizing false positives, a common challenge in malware detection systems. Moreover, the study uses a hybrid approach, combining supervised learning for malware classification with unsupervised learning to identify unknown threats based on behavioral anomalies. Experimental results demonstrate the effectiveness of the proposed approach in detecting novel malware strains that bypass traditional detection methods. The system achieves high detection rates with reduced latency, proving its potential for real-world application in dynamic cybersecurity environments. This research highlights the transformative potential of deep learning data-driven anomaly detection as a critical safeguard against emerging and sophisticated cyber threats.
| Item Type: | Book Chapter |
|---|---|
| Subjects: | Q Science > Q Science (General) > Q300-390 Cybernetics |
| Divisions: | School of Information Technology, Engineering, Mathematics and Physics (STEMP) |
| Depositing User: | Sushita Sharma |
| Date Deposited: | 09 Feb 2026 02:30 |
| Last Modified: | 09 Feb 2026 02:30 |
| URI: | https://repository.usp.ac.fj/id/eprint/15254 |
Actions (login required)
 |
View Item |