@Article{cmes.2025.064926,
AUTHOR = {Madini O. Alassafi, Syed Hamid Hasan},
TITLE = {Dual-Channel Attention Deep Bidirectional Long Short Term Memory for Enhanced Malware Detection and Risk Mitigation},
JOURNAL = {Computer Modeling in Engineering \& Sciences},
VOLUME = {144},
YEAR = {2025},
NUMBER = {2},
PAGES = {2627--2645},
URL = {http://www.techscience.com/CMES/v144n2/63697},
ISSN = {1526-1506},
ABSTRACT = {Over the past few years, Malware attacks have become more and more widespread, posing threats to digital assets throughout the world. Although numerous methods have been developed to detect malicious attacks, these malware detection techniques need to be more efficient in detecting new and progressively sophisticated variants of malware. Therefore, the development of more advanced and accurate techniques is necessary for malware detection. This paper introduces a comprehensive Dual-Channel Attention Deep Bidirectional Long Short-Term Memory (DCA-DBiLSTM) model for malware detection and risk mitigation. The Dual Channel Attention (DCA) mechanism improves the model’s capability to concentrate on the features that are most appropriate in the input data, which reduces the false favourable rates. The Bidirectional Long, Short-Term Memory framework helps capture crucial interdependence from past and future circumstances, which is essential for enhancing the model’s understanding of malware behaviour. As soon as malware is detected, the risk mitigation phase is implemented, which evaluates the severity of each threat and helps mitigate threats earlier. The outcomes of the method demonstrate better accuracy of 98.96%, which outperforms traditional models. It indicates the method detects and mitigates several kinds of malware threats, thereby providing a proactive defence mechanism against the emerging challenges in cybersecurity.},
DOI = {10.32604/cmes.2025.064926}
}