You are required to read and agree to the below before accessing a full-text version of an article in the IDE article repository.

The full-text document you are about to access is subject to national and international copyright laws. In most cases (but not necessarily all) the consequence is that personal use is allowed given that the copyright owner is duly acknowledged and respected. All other use (typically) require an explicit permission (often in writing) by the copyright owner.

For the reports in this repository we specifically note that

  • the use of articles under IEEE copyright is governed by the IEEE copyright policy (available at http://www.ieee.org/web/publications/rights/copyrightpolicy.html)
  • the use of articles under ACM copyright is governed by the ACM copyright policy (available at http://www.acm.org/pubs/copyright_policy/)
  • technical reports and other articles issued by M‰lardalen University is free for personal use. For other use, the explicit consent of the authors is required
  • in other cases, please contact the copyright owner for detailed information

By accepting I agree to acknowledge and respect the rights of the copyright owner of the document I am about to access.

If you are in doubt, feel free to contact webmaster@ide.mdh.se

Unmasking Novel IoT Threats through Attack Disjoint Feature Attribution and Transparent Machine Learning

Authors:

DANJELA KUÇI , Mobyen Uddin Ahmed, Shahid Raza

Research group:

Artificial Intelligence and Intelligent Systems

Publication Type:

Conference/Workshop Paper

Venue:

9th Artificial Intelligence and Cloud Computing Conference

Abstract

This study investigates machine learning and explainable artificial intelligence techniques for detecting previously unseen IoT cyberattacks using an attack‑disjoint evaluation methodology, in which models are trained on one set of attack categories and tested exclusively on withheld attack types. A multi‑model intrusion detection framework—comprising Random Forest (RF), XGBoost, Isolation Forest, One‑Class SVM, and Autoencoder—was evaluated on the CIC-IoT2023 dataset and validated for cross dataset generalisation on the UNSW NB15 dataset. Precision was enhanced through threshold optimisation and cost‑sensitive learning, while ablation studies quantified the impact of class balancing, ensemble size and SHAP‑based feature selection. SHAP and LIME explainability analyses highlighted key influential features and clarified model behaviour. Results show RF achieving the highest recall (90.05%) and XGBoost the highest precision (95%) in the attack‑disjoint setting, with strong generalisation across datasets. Per‑attack‑type evaluation revealed substantial variability, with Backdoor_Malware easiest to detect and SQL Injection most challenging.

Bibtex

@inproceedings{KUCI7450,
author = {DANJELA KU{\c{C}}I and Mobyen Uddin Ahmed and Shahid Raza},
title = {Unmasking Novel IoT Threats through Attack Disjoint Feature Attribution and Transparent Machine Learning},
month = {February},
year = {2027},
booktitle = {9th Artificial Intelligence and Cloud Computing Conference },
url = {http://www.es.mdu.se/publications/7450-}
}