Scholarlink

The proliferation of the Internet of Things (IoT) has revolutionized various industries, offering enhanced connectivity and automation. However, this rapid expansion has also introduced significant security vulnerabilities, making IoT networks prime targets for cyber-attacks. Intrusion Detection Systems (IDS) are crucial for safeguarding these networks, providing real-time detection and response to potential threats. Machine learning-based IDS have emerged as effective solutions, offering the ability to analyze large volumes of network data and identify malicious activities with high accuracy. In this paper, we present a novel IDS model utilizing a Multi-Layer Perceptron (MLP), combined with Random Forest-based feature selection, to optimize detection performance while reducing model complexity. We evaluated the MLP model using four distinct feature sets (10, 20, 30, and 40 features) derived from the Ton-IoT dataset to assess the impact of feature selection on accuracy and efficiency. Subsequently, model pruning was applied to reduce the model size, ensuring its suitability for deployment in resource-constrained IoT environments. Our results demonstrate that the proposed approach effectively balances detection performance with resource efficiency, making it a robust solution for securing IoT networks.

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

Efficient IoT Intrusion Detection: Integrating Random Forest Feature Selection with MLP Model Pruning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kamal Bella,
  • Mouaad Mohy-Eddine,
  • Azidine Guezzaz,
  • Said Benkirane,
  • Mourade Azrour

摘要

The proliferation of the Internet of Things (IoT) has revolutionized various industries, offering enhanced connectivity and automation. However, this rapid expansion has also introduced significant security vulnerabilities, making IoT networks prime targets for cyber-attacks. Intrusion Detection Systems (IDS) are crucial for safeguarding these networks, providing real-time detection and response to potential threats. Machine learning-based IDS have emerged as effective solutions, offering the ability to analyze large volumes of network data and identify malicious activities with high accuracy. In this paper, we present a novel IDS model utilizing a Multi-Layer Perceptron (MLP), combined with Random Forest-based feature selection, to optimize detection performance while reducing model complexity. We evaluated the MLP model using four distinct feature sets (10, 20, 30, and 40 features) derived from the Ton-IoT dataset to assess the impact of feature selection on accuracy and efficiency. Subsequently, model pruning was applied to reduce the model size, ensuring its suitability for deployment in resource-constrained IoT environments. Our results demonstrate that the proposed approach effectively balances detection performance with resource efficiency, making it a robust solution for securing IoT networks.