Scholarlink

Android has become increasingly popular due to the wide availability of applications across multiple app stores. Previous studies have proposed malware detection models based on centralized techniques; however, these approaches raise significant privacy concerns for users. To overcome this limitation, we propose a federated learning-based framework that enables privacy-preserving malware detection. The framework employs a Deep Neural Network (DNN) for training on the cloud side, while a semi- supervised machine learning approach is applied on the client side. Experiments were conducted on a dataset of 2,00,000 Android applications distributed across 200 clients over multiple rounds of federation. Results show that the proposed framework achieves 98.7% accuracy when tested on real-world applications, highlighting its effectiveness and applicability.

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

PermDroid: A Privacy-Preserving Framework for Android Malware Detection Using Federated Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ranjan Manhas,
  • Arvind Mahindru,
  • Himani Arora

摘要

Android has become increasingly popular due to the wide availability of applications across multiple app stores. Previous studies have proposed malware detection models based on centralized techniques; however, these approaches raise significant privacy concerns for users. To overcome this limitation, we propose a federated learning-based framework that enables privacy-preserving malware detection. The framework employs a Deep Neural Network (DNN) for training on the cloud side, while a semi- supervised machine learning approach is applied on the client side. Experiments were conducted on a dataset of 2,00,000 Android applications distributed across 200 clients over multiple rounds of federation. Results show that the proposed framework achieves 98.7% accuracy when tested on real-world applications, highlighting its effectiveness and applicability.