An Artificial Intelligence Driven Optimal Deep Belief Network Model for Malware Classification on IoT-Cloud Environment

Computer-based systems, including mobile devices, desktops, Internet of Things (IoT), and Cyber-Physical Systems (CPS), are designed to protect data effectively. However, malware targets these systems, threatening data accessibility, integrity, and confidentiality through cyberattacks. This study proposes an Artificial Intelligence-driven Optimal Deep Belief Network for Malware Detection and Classification (AIODBN-MDC) approach, aiming to detect and classify malware in IoT-based cloud infrastructure. Initially, z-score normalization is performed to scale the data in a standard form. Then, a Bottleneck-Driven DBN (BDDBN) model is utilized to detect and classify the malware. Finally, the Enhanced Grasshopper Optimization Algorithm (EGOA) model is employed to fine-tune the hyperparameters of the BDDBN classifier. Experimental investigation of the proposed AIODBN-MDC technique on an Android malware dataset demonstrated an accuracy of 99.34%, outperforming existing methods.