Application and challenge analysis of deep learning in malicious software detection
Linxi Wang
The University of New South Wales
DOI: https://doi.org/10.59429/esta.v11i4.8485
Keywords: Deep learning; Malware detection; Adversarial training
Abstract
Deep learning has become a critical tool in detecting malware, offering advanced techniques such as automated feature learning and behavioral analysis. Its ability to handle large datasets and detect sophisticated threats makes it highly effective, especially in scenarios where traditional methods fail. However, challenges remain, including data imbalance, computational demands, and vulnerability to adversarial attacks. Hybrid models, adversarial training, and explainability solutions are being explored to address these limitations. With continued advancements in model efficiency and transparency, deep learning’s role in cybersecurity will become increasingly practical and robust, providing a more adaptive approach to malware detection.
References
[1] Chen Yi, Tang Di, Zou Wei. Android malware detection based on deep learning: achievements and challenges [J]. Journal of Electronics and Information Technology, 2020, 42 (9): 13.
[2] Wei Gaoshan, Yu Shugang, Xian Jiemin, et al. Research on Malicious Software Detection Technology Based on Deep Learning [J]. Information Technology, 2022 (009): 046.
[3] Wu Xiaomei. Research on the Application of Deep Learning in Malicious Software Analysis [J]. Computer Application Digest, 2023, 39 (21): 40-42.
[4] Zhang Hao. Research on Dynamic Detection Method of Malicious Software Based on Deep Learning [J]. Electronic Technology and Software Engineering, 2022 (003): 000.