Scholarlink

The rapid growth of digital content and connected devices has led to unprecedented data flow demands across global networks, giving rise to new cybersecurity challenges. The increasing frequency of data breaches and the proliferation of diverse malware, particularly targeting IoT devices, underscore the urgent need for advanced security solutions to address the evolving threat landscape. This paper introduces an advanced malware detection framework that integrates Federated Learning with Trend Micro Locality Sensitive Hashing (TLSH) clustering to enhance malware detection. By leveraging the decentralized nature of Federated Learning and the clustering capabilities of TLSH, the proposed method improves detection accuracy while preserving data privacy. Experimental evaluations on the Tritium and Inferno malware datasets demonstrate the model’s ability to deliver highly accurate malware classification, making it scalable for a wide range of contemporary cybersecurity environments.

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

Comprehensive Framework for Malware Detection Using Federated Learning with TLSH-Based Clustering

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Vinayak Anand,
  • Kamalakanta Sethi

摘要

The rapid growth of digital content and connected devices has led to unprecedented data flow demands across global networks, giving rise to new cybersecurity challenges. The increasing frequency of data breaches and the proliferation of diverse malware, particularly targeting IoT devices, underscore the urgent need for advanced security solutions to address the evolving threat landscape. This paper introduces an advanced malware detection framework that integrates Federated Learning with Trend Micro Locality Sensitive Hashing (TLSH) clustering to enhance malware detection. By leveraging the decentralized nature of Federated Learning and the clustering capabilities of TLSH, the proposed method improves detection accuracy while preserving data privacy. Experimental evaluations on the Tritium and Inferno malware datasets demonstrate the model’s ability to deliver highly accurate malware classification, making it scalable for a wide range of contemporary cybersecurity environments.