Sign in Subscribe
Machine learning

Advanced Machine Learning Models Enhance IoT Security

Strategic Insights

— 2 min read

Investigations into the performance of artificial intelligence-based intrusion detection systems in the Internet of Things (IoT) environment have demonstrated significant improvements in cybersecurity. Research conducted by the University of Information Technology and Communications has shown that advanced Machine Learning (ML) and Deep Learning (DL) models can achieve high accuracy rates in detecting malicious activities, such as botnets and Distributed Denial-of-Service (DDoS) attacks. These findings suggest a significant leap forward in safeguarding IoT devices against cyber threats, with real-time and scalable intrusion detection capabilities becoming a viable option.

Key Takeaways:

  • Advanced ML models, including XGBoost, have demonstrated superior performance in detecting malicious activities on IoT devices, achieving 99.99% accuracy on the BoT-IoT dataset.
  • Deep Learning models, specifically Convolutional Neural Networks (CNNs), have achieved 99.99% accuracy on the BoT-IoT dataset with preprocessing, highlighting the critical role of data preparation.
  • The study also found that preprocessing techniques play a crucial role in enhancing IoT security, providing a pathway for real-time, scalable intrusion detection in IoT environments.
  • The research used two benchmark datasets, BoT-IoT and CIC-IDS2017, to evaluate the performance of ML and DL models.
  • XGBoost demonstrated the best performance among ML models, achieving 99.91% accuracy on the CIC-IDS2017 dataset.
  • Convolutional Neural Networks (CNNs) achieved 99.61% accuracy on the CIC-IDS2017 dataset with preprocessing.
  • The study concludes that advanced ML/DL models and preprocessing techniques can effectively enhance IoT security, providing a pathway for real-time, scalable intrusion detection in IoT environments.
  • The research was conducted by the University of Information Technology and Communications and published in the Iraqi Journal for Computers and Informatics.
  • The authors of the study are Karrar Majid Jasim from the University of Information Technology and Communications.

Statistics:

  • 99.99% accuracy achieved by XGBoost on the BoT-IoT dataset.
  • 99.91% accuracy achieved by XGBoost on the CIC-IDS2017 dataset.
  • 99.99% accuracy achieved by Convolutional Neural Networks (CNNs) on the BoT-IoT dataset with preprocessing.
  • 99.61% accuracy achieved by CNNs on the CIC-IDS2017 dataset with preprocessing.
  • The research used two benchmark datasets: BoT-IoT and CIC-IDS2017.
  • The study concluded that advanced ML/DL models can effectively enhance IoT security, providing a pathway for real-time, scalable intrusion detection in IoT environments.

Sources:

  • Iraqi Journal for Computers and Informatics, 2025,51(1):83-93.
  • Iraqi Journal for Computers and Informatics - http://www.uoitc.edu.iq/ijci1/index.html.
  • University of Information Technology and Communications.
  • karrar Majid Jasim, University of Information Technology and Communications.

Read more