Scholarlink

The present paper focuses on deep learning-enabled modelling to find the abnormalities in near-time cloud data and categorize the multiple network assaults. The near real-time UNSW-NB15 information sets were utilized to design and develop the deep convolutional neural network (CNN) model applied to the best twelve features. The decision tree-based random forest algorithm has been applied to the preprocessed UNSW-NB15 data to obtain the appropriate features for the deep CNN model. The experimental results of the deep CNN model with twelve features were assessed with 93.82% prediction accuracy on the validation data with negligible loss. Thus, the proposed approach in selecting features and detecting abnormalities in the network is superior and is the best choice for network intrusion detection in real time.

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

Multiple Intrusion Detection in Complex Cloud Environments Using Random Forest and Deep Learning on the UNSW-NB15 Benchmark Datasets

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Abhinav Upadhyay,
  • Nidhi Thakur,
  • Abhishek Pandey,
  • Minhaj Khan,
  • Amol D. Vibhute

摘要

The present paper focuses on deep learning-enabled modelling to find the abnormalities in near-time cloud data and categorize the multiple network assaults. The near real-time UNSW-NB15 information sets were utilized to design and develop the deep convolutional neural network (CNN) model applied to the best twelve features. The decision tree-based random forest algorithm has been applied to the preprocessed UNSW-NB15 data to obtain the appropriate features for the deep CNN model. The experimental results of the deep CNN model with twelve features were assessed with 93.82% prediction accuracy on the validation data with negligible loss. Thus, the proposed approach in selecting features and detecting abnormalities in the network is superior and is the best choice for network intrusion detection in real time.