Related Experiment Videos
AHGA-SA: A Novel Adaptive Hybrid Framework for Feature Selection in IoT-Oriented Intrusion Detection with Explainable
Saud Abdullah Alzughaibi1, Iftikhar Ahmad1, Madini Alassafi1
1Department of Information Technology, Faculty of Computing and Information Technology, King Abdulaziz University, Jeddah 21589, Saudi Arabia.
Sensors (Basel, Switzerland)
|July 15, 2026
Summary
This study introduces an Adaptive Hybrid Genetic Algorithm-Simulated Annealing (AHGA-SA) for efficient feature selection in intrusion detection systems (IDS). AHGA-SA significantly reduces data complexity and computational costs while maintaining high detection accuracy for IoT environments.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Machine Learning
Background:
- Internet of Things (IoT) environments face increasing cyberattack risks.
- Intrusion Detection Systems (IDS) are crucial for IoT security.
- Effective feature selection (FS) in IDS enhances performance and reduces complexity.
Purpose of the Study:
- To present the Adaptive Hybrid Genetic Algorithm-Simulated Annealing (AHGA-SA) as a novel FS framework.
- To optimize feature subsets for high-dimensional intrusion-detection datasets.
- To maintain high detection performance while reducing computational cost.
Main Methods:
- Integration of genetic algorithm (global search) and simulated annealing (local exploitation).
- Application of AHGA-SA for feature selection in intrusion detection.
- Utilizing Shapley additive explanations (XAI) for model interpretability.
Main Results:
- Achieved high classification accuracies: 99.04% (IoTID20), 98.25% (WUSTL-EHMS), 99.18% (Edge-IIoTset).
- Demonstrated significant feature-space reduction (12, 7, and 9 features respectively).
- Reduced training/testing times, CPU usage, memory overhead, and subset size compared to baseline methods.
Conclusions:
- AHGA-SA effectively identifies compact and informative feature subsets for IDS.
- The framework enhances IDS performance and efficiency in IoT security.
- Explainable AI confirms the contribution of selected features to IDS decisions.