Interpretable intrusion detection for IoT environments using a self-attention-based explainable AI framework
- Kanta Prasad Sharma 1, Tapsi Nagpal 2, Tarak Vora 3, Anupam Yadav 4, Muhammad Irsyad Abdullah 5, B Jayaprakash 6, Aditya Kashyap 7,8, G Sridevi 9, A Bhowmik 10,11, Bethelehem Burju Bukate 12
- 1Computer Science & Engineering, Amity University Greater Noida Campus, Noida, India.
- 2Department of Computer Science and Engineering, Lingaya's Vidyapeeth, Faridabad, India.
- 3Department of Civil Engineering, Faculty of Engineering & Technology, Marwadi University Research Center, Marwadi University, Rajkot, Gujarat, India.
- 4Department of Computer Engineering and Application, GLA University, Mathura, 281406, India.
- 5Management and Science University, Shah Alam, Selangor, Malaysia.
- 6Department of Computer Science & IT, School of Sciences, JAIN (Deemed to Be University), Bangalore, Karnataka, India.
- 7Centre for Research Impact & Outcome, Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura, Punjab, 140401, India.
- 8Sharda School of Engineering and Sciences, Sharda University, Knowledge Park III, Greater Noida, Uttar Pradesh, 201310, India.
- 9Department of CSE, Raghu Engineering College, Visakhapatnam, Andhra Pradesh, 531162, India.
- 10Department of Additive Manufacturing, Mechanical Engineering, SIMATS, Saveetha Institute of Medical and Technical Sciences, Thandalam, Chennai, 602105, India.
- 11Division of Research and Development, Lovely Professional University, Phagwara, Punjab, India.
- 12Faculty of Mechanical Engineering, Jimma Institute of Technology, Jimma University, Jimma, Ethiopia. bethelhem.burju@ju.edu.et.
- 0Computer Science & Engineering, Amity University Greater Noida Campus, Noida, India.
Related Experiment Videos
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study introduces a novel self-attention deep neural network with learnable feature gating for enhanced Internet of Things (IoT) cybersecurity. The model achieves high accuracy in detecting network intrusions, offering a robust and interpretable solution.
Area Of Science
- Cybersecurity
- Artificial Intelligence
- Network Security
Background
- The proliferation of Internet of Things (IoT) devices presents significant cybersecurity challenges due to expanded attack surfaces.
- Existing intrusion detection systems often struggle with the complexity and scale of modern network traffic.
Purpose Of The Study
- To propose a novel self-attention deep neural network (SA-DNN) integrated with a learnable feature gating (LFG) mechanism for robust IoT intrusion detection.
- To enhance model adaptability and feature optimization without manual feature selection.
Main Methods
- Development of a SA-DNN architecture incorporating an LFG mechanism for dynamic feature emphasis.
- Utilizing self-attention to capture long-range dependencies in network traffic.
- Employing SHAP and LIME for model interpretability.
Main Results
- Achieved 99.3% accuracy on the BoT-IoT dataset and 99.6% on the N-BaIoT dataset.
- Demonstrated strong generalizability with 97.9% accuracy on the UNSW-NB15 dataset.
- Outperformed baseline and state-of-the-art methods in intrusion detection.
Conclusions
- The proposed SA-DNN with LFG offers a lightweight, scalable, and interpretable solution for IoT cybersecurity.
- The model exhibits robustness and strong generalizability across diverse network intrusion scenarios.
- This approach enhances transparency in intrusion detection decisions.
Related Experiment Videos
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.