Intelligent machine learning solutions with Bayesian regularization backpropagation adaptive networks for differential systems of Maize streak virus diseases.

Maize streak virus, a cause of maize streak disease, poses a major threat to food security that significantly reduces yields of maize, a crucial food crop, thereby jeopardizing the volatile social and economic stability of subsistence growers across the planted field. This communication emphasizes the predictive forecasting of epidemic dynamics by including three control interventions: chemical control, quarantine, and preventive, together with the parameter estimations. The study integrates machine predictive neural networks with Bayesian regularization (NNBNRS) backpropagation for an efficient oversight mechanism for predicting the dynamic behavior of maize streak virus disease (MSVD). The four compartments governing the MSVD non-linear differential framework constitute the healthy and infected maize as well as the healthy and infected leafhoppers. Synthetic datasets are generated for the construction of NNBNRS by means of the numerical explicit Runge-Kutta solver by incorporating the adaptation based on the carrying capacity, the impact of control strategies, the half-saturation ratio of healthy leafhoppers with infectious maize, and the infection and predation rate of healthy leafhoppers on infectious maize. The NNBNRS segmented the synthetic datasets into the testing and training samples to determine the machine predictive solutions for non-linear MSVD systems. The efficacy achieved on fitness performance via mean squared errors, exhaustive regression assessments, and error histograms, is utilized to substantiate the convergence, precision, and reliability of the proposed NNBNRS for a non-linear MSVD system.