Taylor Optimal Strategy with Starling Murmuration Optimizer and Kolmogorov-Arnold Networks for Detect Security Attack in IoT Environment

Internet of Things (IoT) is a rapidly advancing technology with diverse applications including cybersecurity and Intrusion Detection Systems (IDS).As IoT networks grow, ensuring a secure infrastructure has become a critical requirement.However, the proliferation of interconnected and remotely accessible smart devices has significantly heightened security challenges.The existing research models struggles to select relevant features and address insufficient security measures for IoT growth and transmitting sensitive information through traditional methods that impact classification accuracy.This research proposes Taylor Optimal Strategy with Starling Murmuration Optimizer (TOSSMO) and Kolmogorov-Arnold Networks (KAN) for network security.The key innovation lies in TOSSMO's bio-inspired optimization for feature selection and KAN's adaptive learning for robust classification and anomaly detection in cybersecurity.The data used in this research is collected from ToN-IoT, BoT-IoT, NSL-KDD and CIC-IDS-2018 datasets, which are then pre-processed using Min-Max normalization for standardizing the feature vectors.Feature selection was carried out using the TOSSMO algorithm which efficiently identifies relevant features while dynamically balancing exploration and exploitation through TOS.The model employed classification using KAN which enables the analysis of various attacks in IDS by efficiently mapping non-linear features and addressing highdimensionality issues.The proposed method achieves a better accuracy 99.98% on ToN-IoT, and 99.99% accuracy on the BoT-IoT dataset, outperforming existing Graph Neural Network (GNN) technique.