Scholarlink

<p>Android malware is increasingly becoming a serious security threat to users as the popularity of mobile phones continues to rise. The application of machine learning models for classifying Android malware has been widely used in related studies. However, machine learning models can outperform techniques utilizing Generative Adversarial Networks (GANs). This study not only experiments with the application of Android malware classification techniques using machine learning algorithms ranging from traditional methods to advanced approaches, including Random Forest (RF), Deep Neural Network (DNN), Extra Trees (ET), LightGBM, and Convolutional Neural Network (CNN), but also employs the adversarially trained Auxiliary Classifier Generative Adversarial Network (AC-GAN) to analyze their effectiveness in classifying Android malware. The study uses the Drebin and CIC-MalDroid2020 datasets for experimentation. The research yielded high accuracy results, with AC-GAN achieving the highest accuracy of 99.73% on the Drebin dataset, and LightGBM achieving the highest accuracy of 98.69% on the CIC-MalDroid2020 dataset. These results demonstrate competitive performance compared to recent studies, with low false positive rates. The findings of this research can serve as a reference for future related studies.</p>

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

Enhanced android malware classification using multi machine learning models and generative adversarial network

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Nguyen Tan Cam,
  • Nguyen Cong Danh,
  • Nghi Hoang Khoa

摘要

Android malware is increasingly becoming a serious security threat to users as the popularity of mobile phones continues to rise. The application of machine learning models for classifying Android malware has been widely used in related studies. However, machine learning models can outperform techniques utilizing Generative Adversarial Networks (GANs). This study not only experiments with the application of Android malware classification techniques using machine learning algorithms ranging from traditional methods to advanced approaches, including Random Forest (RF), Deep Neural Network (DNN), Extra Trees (ET), LightGBM, and Convolutional Neural Network (CNN), but also employs the adversarially trained Auxiliary Classifier Generative Adversarial Network (AC-GAN) to analyze their effectiveness in classifying Android malware. The study uses the Drebin and CIC-MalDroid2020 datasets for experimentation. The research yielded high accuracy results, with AC-GAN achieving the highest accuracy of 99.73% on the Drebin dataset, and LightGBM achieving the highest accuracy of 98.69% on the CIC-MalDroid2020 dataset. These results demonstrate competitive performance compared to recent studies, with low false positive rates. The findings of this research can serve as a reference for future related studies.