Lightweight IoT IDS keeps accuracy with three to 15 features
Original title: Towards Lightweight IoT Intrusion Detection: Feature Optimization and Explainable Machine Learning for Resource-Constrained Environment
On CICIoT2023, three features reached MCC 0.9622 and 15 preserved 99.97% of a 38-feature baseline.
A paper proposes feature optimization for machine-learning intrusion detection on resource-constrained IoT devices, using CICIoT2023. Random Forest ranking selected compact subsets, and SHAP was used to interpret decisions. A six-feature model reached MCC 0.9675 after an 84.2% feature reduction, with only a small drop from the full model.
- A three-feature model cut the feature space by 92.1%.
- The 15-feature model reached MCC 0.9744 with a 60.5% reduction.
- The full baseline used 38 features.
- SHAP analysis was used to check the selected features.
From the original article
1 min read at arxiv.orgMachine learning-based intrusion detection systems (IDSs) can improve IoT security but often rely on large feature sets that increase computational overhead and limit deployment on resource-constrained devices. This paper presents a resource-efficient and explainable feature optimization framework for lightweight IoT intrusion detection using the CICIoT2023 dataset.
By Shaker Nawasra · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.