MCWOA based Hybrid Deep Learning for Detecting the Attacks in Cybersecurity with IoT Network

All throughout the globe, machine learning is being used in a diversity of settings, from universities to businesses. Intuitive machine learning algorithms can assess dangers and react quickly to security breaches and problems. In the realm of cybersecurity, it is essential for providing a proactive security system. Cybersecurity safeguards data, systems, and networks against unauthorized access in real time. Multiple analyses of privacy and security metrics over the past decade have found that the frequency and severity of cybersecurity breaches have been steadily increasing. Criminals are penetrating data protection systems at a rapid clip. Top cyber-security problems that need efficient explanations include anomaly finding, software vulnerability diagnosis, phishing page proof of identity, malware identification. In addition, NSL-KDD or KDD-CUP99 datasets provide the basis of most extant research. None of these datasets contain assaults that occurred recently. Networkflow ToN-IoT, a realistic dataset obtained from a large-scale, heterogeneous IoT network, was therefore utilized in the study. The study used a hybrid deep learning model combining ResNet and AlexNet for classification purposes. This study employs a hybrid optimization method that takes cues from both the Chimp Optimization Algorithm (COA) and the Whale Optimization Algorithm (WOA) to fine-tune AlexNet's parameters. In order to recover the accuracy of MCWOA's identification, the Sobol arrangement is employed during the initialization of the population to distribute it uniformly over the solution space. Additionally, the position update process of classic MCWOA is enhanced by combining the bubble-net hunting and random search processes of the whale optimization algorithm. The results have important consequences for the continuous creation of adaptable, efficient, and secure intrusion detection systems for the complicated environment of IIoT networks Hybrid (ResNet-AlexNet) model reached the accuracy as 99.43 besides precision of 98.91 besides recall as 97.54 and the f1-score as 99.28 congruently.