Quantum-Integrated Adjustable Botox Neural Network-based Laplacian of Gaussian Framework for Advanced Internet of Things and Network Intrusion Detection

The rapid growth of interconnected systems and Internet of Things (IoT) devices has significantly increased the vulnerability of networks to sophisticated cyberattacks. Traditional anomaly detection approaches, though effective in certain scenarios, often struggle with high-dimensional data, non-linear traffic patterns, and evolving attack strategies. This research proposes a hybrid quantum-classical framework that uses Quantum Radial Basis Function (QRBF) kernels for enhanced feature representation, coupled with an Adjustable Botox Neural Network-based Laplacian of Gaussian (ANNLoG) classifier for precise anomaly detection. Preprocessing is performed using Min-Max normalization to standardize multi-source datasets, including UNSW-NB15, CIC-IDS 2017, and ToN-IoT. The Botox Optimization Algorithm is employed for hyper parameter tuning of ANNLoG. The proposed ANNLoG model achieved exceptional performance, attaining $99.9 \%, 99.8 \%$, and 99.7% accuracy on UNSW-NB15, CIC-IDS 2017, and ToN-IoT datasets, respectively, ensuring superior precision, recall, and $\mathbf{F 1}$-scores. The ANNLoG model demonstrated minimal errors, achieving MSE, MAE, and RMSE of 0.0010,0.0080, and 0.0316 on CICIDS 2017, with similarly low values for UNSW-NB15 and ToNIoT datasets.