Scholarlink

The widespread adoption of Android applications has led to a surge in malware threats, necessitating efficient and scalable detection techniques. This study presents a hybrid optimization framework that combines Harris-Hawks Optimization (HHO) for feature selection with Differential Evolution (DE) for classifier parameter tuning. Mutual information filtering and ADASYN have been used by this framework to increase feature connection and class balance. A soft voting mechanism using SVM, Random Forest, and XGBoost has been applied for the final classification. By achieving an accuracy of 98.26%, the model surpasses traditional models. In the end, the results highlight how well metaheuristic optimization and ensemble learning work together. Live detection and adversity strength are two possible time-ahead enhancements.

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

A Hybrid Optimization Approach for Android Malware Detection: Integrating Harris Hawk Optimization and Differential Evolution

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Nisha Rajawat,
  • Bam Bahadur Sinha

摘要

The widespread adoption of Android applications has led to a surge in malware threats, necessitating efficient and scalable detection techniques. This study presents a hybrid optimization framework that combines Harris-Hawks Optimization (HHO) for feature selection with Differential Evolution (DE) for classifier parameter tuning. Mutual information filtering and ADASYN have been used by this framework to increase feature connection and class balance. A soft voting mechanism using SVM, Random Forest, and XGBoost has been applied for the final classification. By achieving an accuracy of 98.26%, the model surpasses traditional models. In the end, the results highlight how well metaheuristic optimization and ensemble learning work together. Live detection and adversity strength are two possible time-ahead enhancements.