Scholarlink

In recent years, malware targeting Android has become a significant threat to users. This paper proposes an accurate detection method that utilizes a deep learning-based stacking approach to enhance robustness against evolved Android malware. When using a non-time-series dataset, the standalone deep learning model achieved an accuracy of 98.27%, while the proposed model reached 99.12%, an improvement of just 0.85%. However, when using a time-series dataset, the standalone deep learning model’s accuracy remained at 93.99%, whereas the proposed model significantly outperformed it, achieving 99.28%. This represents a substantial accuracy improvement of 5.29%, clearly demonstrating the enhanced robustness of the proposed model.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Accurate and Robust Deep Learning-Based Stacked Generalization Method for Android Malware Detection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ayumu Masudome,
  • Tao Ban,
  • Takeshi Takahashi,
  • Tsung-Nan Lin,
  • Tomohiro Morikawa

摘要

In recent years, malware targeting Android has become a significant threat to users. This paper proposes an accurate detection method that utilizes a deep learning-based stacking approach to enhance robustness against evolved Android malware. When using a non-time-series dataset, the standalone deep learning model achieved an accuracy of 98.27%, while the proposed model reached 99.12%, an improvement of just 0.85%. However, when using a time-series dataset, the standalone deep learning model’s accuracy remained at 93.99%, whereas the proposed model significantly outperformed it, achieving 99.28%. This represents a substantial accuracy improvement of 5.29%, clearly demonstrating the enhanced robustness of the proposed model.