Design of backpropagated predictive computing network for computer virus propagation model with countermeasures

This study introduces the novel backpropagated heuristics computing paradigm (BHCP) to an epidemic model, specifically demonstrating effectiveness in modeling computer virus propagation (CVP) under supervised control. It is based on the design of artificial neural networks (ANNs) optimized using the efficient Levenberg-Marquardt scheme (LMS), i.e., ANNs-LMS. The non-linear CVP model classifies computer interactions into susceptible, infected, and countermeasure categories, for studying internal and external computer interactions in cybersecurity. The Adams method is used to create a synthetic dataset with varying rates of infection, reinfection, recovery, and disconnection of infected computers, and countermeasure propagation and infection recurrence across distinct situations. The proposed results are compared with referenced solutions to demonstrate the significance of the designed BHCP for solving the CVP model in terms of complexity, accuracy, and convergence. The efficacy and legacy of the designed approach are validated using mean squared error performance, fitness evaluations, error histogram analysis, and regression metrics.