ASEAN Journal on Science and Technology for Development
Abstract
The rising prevalence of obfuscated malware poses a critical threat to network security, undermining traditional detection methods and jeopardizing data integrity and system reliability in an increasingly connected world. This growing danger highlights the urgent need for advanced solutions to protect against evolving cyber risks. This research intro-duces a novel framework to enhance malware detection, employing a hybrid architecture that integrates spatial and temporal analysis with attention mechanisms. The approach leverages a large dataset subsample, focusing on key feature selection and augmentation to improve robustness against evasion techniques. This innovative framework offers a significant advancement in identifying malicious network traffic, addressing a vital gap in current security practices. The result demonstrates the model’s effectiveness, achieving an accuracy of 97.22%, underscoring its potential to strengthen defenses against sophisticated threats. Future efforts could further refine its performance, reinforcing its role in safeguarding network environments.
Keywords
Malware Detection; Network Security; Deep Learning; Obfuscation; Threat Detection
Publication Date
2026
Received Date
16 Oct 2025
Revised Date
24 Feb 2026
Accepted Date
14 Mar 2026
Recommended Citation
Parmar, Meera and Gautam, Sunil
(2026)
"A Deep Neural Approach to Network-Based Obfuscated Malware,"
ASEAN Journal on Science and Technology for Development: Vol. 43:
No.
2, Article 2.
DOI: https://doi.org/10.61931/ajstd.2224-9028.1657
Available at:
https://ajstd.ubd.edu.bn/journal/vol43/iss2/2