Related Experiment Video
Updated: Sep 17, 2025
07:35
Selecting Multiple Biomarker Subsets with Similarly Effective Binary Classification Performances
Published on: October 11, 2018
7.6K
Integration of metaheuristic based feature selection with ensemble representation learning models for privacy aware
M Karthikeyan1, R Brindha1, Maria Manuel Vianny2
1Department of Computing Technologies, School of Computing, SRM Institute of Science and Technology, Chennai, 603203, India.
Scientific Reports
|July 2, 2025
Summary
This study introduces an advanced AI model for detecting cyberattacks in the Internet of Things (IoT). The Adaptive Metaheuristic-Based Feature Selection with Ensemble Learning Model for Privacy-Preserving Cyberattack Detection (AMFS-ELPPCD) achieves high accuracy in identifying threats.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Internet of Things (IoT)
Background:
- The proliferation of smart devices in the Internet of Things (IoT) increases vulnerability to sophisticated cyberattacks.
- Existing intrusion detection systems (IDS) struggle to keep pace with the complexity of modern cyber threats in IoT environments.
- Advanced AI techniques, including machine learning (ML) and deep learning (DL), offer significant potential for enhancing IDS performance.
Purpose of the Study:
- To propose an novel technique, the Adaptive Metaheuristic-Based Feature Selection with Ensemble Learning Model for Privacy-Preserving Cyberattack Detection (AMFS-ELPPCD), for robust cyberattack detection in IoT.
- To leverage AI for improved accuracy and efficiency in identifying malicious network activities within IoT ecosystems.
- To enhance the privacy-preserving capabilities of intrusion detection systems.
Main Methods:
- Data preprocessing using Z-score normalization.
- Feature selection employing the adaptive Harris hawk optimization (AHHO) model.
- Ensemble classification utilizing bidirectional gated recurrent unit (BiGRU), Wasserstein autoencoder (WAE), and deep belief network (DBN).
- Hyperparameter optimization of ensemble classifiers via social group optimization (SGO).
Main Results:
- The AMFS-ELPPCD technique demonstrated superior performance on benchmark datasets.
- Achieved high accuracy rates of 99.44% on the CICIDS-2017 dataset and 98.85% on the NSLKDD dataset.
- Outperformed existing models in detecting and classifying cyberattacks in IoT environments.
Conclusions:
- The proposed AMFS-ELPPCD model offers a highly effective solution for privacy-preserving cyberattack detection in IoT.
- The integration of metaheuristic optimization and ensemble learning significantly boosts intrusion detection capabilities.
- The findings highlight the potential of advanced AI techniques in securing the rapidly expanding IoT landscape.