Scholarlink

Conventional malware detection is challenged by sophisticated threats employing obfuscation, polymorphism, and sandbox evasion. We introduce the Dynamic Malware Defense Algorithm (DMDA), a sandbox-based system that integrates behavioral analysis, machine learning, and Explainable AI (LIME, SHAP). Hosted on AWS, DMDA operates at 98.8% accuracy with < 1 s delay, better than solutions such as Cuckoo Sandbox and Joe Sandbox ML. With continuous and reinforcement learning, DMDA provides a scalable, adaptive, and explainable real-time solution for contemporary cybersecurity.

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

DeepXDetect: Real-Time Sandbox Malware Detection with Explainable AI

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shaik Salma Begum,
  • G. Suresh Babu,
  • P. C. Senthil Mahesh,
  • Ankita Bhaumik,
  • Saptarsi Sanyal,
  • S. Fayaz Begum

摘要

Conventional malware detection is challenged by sophisticated threats employing obfuscation, polymorphism, and sandbox evasion. We introduce the Dynamic Malware Defense Algorithm (DMDA), a sandbox-based system that integrates behavioral analysis, machine learning, and Explainable AI (LIME, SHAP). Hosted on AWS, DMDA operates at 98.8% accuracy with < 1 s delay, better than solutions such as Cuckoo Sandbox and Joe Sandbox ML. With continuous and reinforcement learning, DMDA provides a scalable, adaptive, and explainable real-time solution for contemporary cybersecurity.