Related Experiment Video
Updated: Jun 21, 2026
06:37
Artificial Intelligence-Based System for Detecting Attention Levels in Students
Published on: December 15, 2023
Attention integrated deep learning models for interpretable multi-class IoT intrusion detection using SHAP.
Ramakrishnan Raman1, Rahul Kumar2, Benson Edwin Raj3
1Higher Colleges of Technology - Dubai Men's Campus, Dubai, United Arab Emirates.
Frontiers in Artificial Intelligence
|June 19, 2026
Summary
This study introduces an attention-enhanced deep learning model for Internet of Things (IoT) intrusion detection, achieving high accuracy in identifying diverse cyber threats. The model effectively combines spatial and temporal analysis for robust network security.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Network Security
Background:
- The Internet of Things (IoT) generates vast, complex network traffic, increasing vulnerability to cyberattacks.
- Traditional intrusion detection systems (IDSs) struggle with novel and sophisticated attack patterns.
- Deep learning offers potential for enhanced detection of complex threats in IoT environments.
Purpose of the Study:
- To develop and evaluate an attention-enhanced deep learning framework for accurate and efficient multi-class intrusion detection in IoT networks.
- To combine spatial feature extraction with temporal dependency analysis for robust anomaly identification.
- To improve the reliability and performance of intrusion detection systems for real-time IoT security.
Main Methods:
- Utilized attention-enhanced one-dimensional Convolutional Neural Network (1D CNN) for spatial feature extraction.
- Employed Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) for analyzing temporal dependencies in network traffic.
- Evaluated the model on the Hacking and Countermeasure Research Lab (HCRL) and Kitsune IoT datasets.
- Applied SHapley Additive exPlanations (SHAP) for model interpretability.
Main Results:
- The attention-enhanced 1D CNN achieved 98% accuracy on the Kitsune dataset and 87% on the HCRL dataset.
- Attention mechanisms significantly improved the model's discriminative capability for complex IoT attack types.
- SHAP analysis identified key features driving intrusion detection decisions, enhancing transparency.
Conclusions:
- The proposed deep learning approach effectively addresses challenges in IoT intrusion detection by integrating spatial and temporal analysis.
- Attention mechanisms are crucial for enhancing the performance of deep learning models in identifying sophisticated cyber threats.
- The framework is suitable for deployment in intelligent, real-time IoT network security systems.