A Comprehensive Study on Capturing Complex Feature Interactions with Advanced Feature Selection and Extraction in IoT Intrusion Detection Systems

Efficient Intrusion Detection Systems (IDS) are required due to the growing cybersecurity threats that have been caused by the Internet of Things’ (IoT) devices. Class imbalance, high-dimensional data, and feature redundancy are common challenges for traditional IDS, which affect detection accuracy and computational complexity. An improved feature selection and extraction method that enhances intrusion detection in Internet of Things networks is presented in this paper. Preprocessing, correlation-based feature selection, autoencoderbased feature extraction, and Synthetic Minority Over-sampling Technique (SMOTE) for balancing datasets are all part of the methodology. A better set of features is examined via a Gradient Boosting classifier that is less computationally expensive with high accuracy. The results show that IDS performance is enhanced through sensitive feature selection, thus qualifying for real-time IoT security applications.