Scholarlink

<p>Although cloud computing networks hold momentous operational efficiency, they face major problems regarding data privacy and the requirement to counter intricate cyber threats, including ransomware. This study introduces an adaptive ransomware detection network that enlists a GELU-activated Recurrent Neural Network (GELU-RNN) abutting Adaptive Hellinger Distance for feature refinement for ransomware detections. The suggested framework emphasizes sustaining privacy via homomorphic encryption while acquiring high detection accuracy and reducing FP. The model successfully identifies intricate and evolving ransomware variants employing identifiers like kurtosis, entropy, and rolling window statistics. The proposed model achieves a higher efficacy, attaining an accuracy of 99.12%, precision of 99.56%, and recall of 98%. A comparative exploration with current techniques highlights the benefits of the suggested approach in tackling ransomware variants while securing sensitive user information.</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

Adaptive Hellinger Distance CS and GELU-RNN: Advanced Solutions for Feature Optimization and Cloud Threat Detection

  • 期刊论文

  • 发表日期：

    2025年6月3日

查看全文
  • Venkata Surya Teja Gollapalli,
  • Rahul Jadon,
  • Rajababu Budda,
  • Kannan Srinivasan,
  • Guman Singh Chauhan,
  • Joseph Bamidele Awotunde

摘要

Although cloud computing networks hold momentous operational efficiency, they face major problems regarding data privacy and the requirement to counter intricate cyber threats, including ransomware. This study introduces an adaptive ransomware detection network that enlists a GELU-activated Recurrent Neural Network (GELU-RNN) abutting Adaptive Hellinger Distance for feature refinement for ransomware detections. The suggested framework emphasizes sustaining privacy via homomorphic encryption while acquiring high detection accuracy and reducing FP. The model successfully identifies intricate and evolving ransomware variants employing identifiers like kurtosis, entropy, and rolling window statistics. The proposed model achieves a higher efficacy, attaining an accuracy of 99.12%, precision of 99.56%, and recall of 98%. A comparative exploration with current techniques highlights the benefits of the suggested approach in tackling ransomware variants while securing sensitive user information.