A Hybrid Autoencoder and Gated Recurrent Unit Model Optimized by Honey Badger Algorithm for Enhanced Cyber Threat Detection in IoT Networks

ABSTRACT The widespread adoption of Internet of Things (IoT) devices has introduced significant security challenges, with cyber threats becoming increasingly sophisticated. This paper proposes a novel approach combining an autoencoder (AE) with a gated recurrent unit (GRU), further enhanced by the Honey Badger Algorithm (HBA) for improved cyber threat detection in IoT environments. The AE, an unsupervised learning technique, is employed to reduce the dimensionality of the data while preserving essential features, making it effective for identifying anomalies linked to cyber threats. The GRU is used to detect complex patterns within the data and recognize various attack types. The HBA optimizes the parameters of the GRU, enhancing its performance in detecting threats. The proposed AE‐GRU‐HBA model is tested on the CSE‐CIC‐IDS2018 dataset, which contains diverse cyber threats. The results indicate that the AE‐GRU model significantly outperforms traditional anomaly detection techniques, offering superior accuracy, faster detection, and greater scalability. This research provides a robust, scalable, and reliable solution for securing IoT networks, ensuring effective defense against evolving cyber threats. Experimental results on the CSE–CIC–IDS2018 dataset show that the model achieves 98.5% accuracy, 98.1% precision, 97.9% recall, and a 98.8% AUC score, outperforming conventional deep learning‐based IDS methods.