Neural Network-Based Numerical Analysis of Maize Streak Epidemic Model with Linear Control

Objectives: The aim of the present study is to conduct a numerical analysis of the performance of various neural architectures in solving the maize streak epidemic model with linear control. A supervised neural network approach is employed to obtain solutions for the maize streak epidemic under linear control. The mathematical model categorizes the maize population into two groups—susceptible and infected—and the vector population into susceptible and infected groups. Method: We compute the numerical solutions of the maize streak epidemic model with linear control using a proficient Bayesian regularization neural network, which is known for effectively solving nonlinear problems. The dataset is generated using the Adams method, with input values ranging from 0 to 1 and a step size of 0.01. The division into training, validation, and testing sets follows appropriate proportions to minimize the mean square error. During training, we apply Bayesian regularization with varying numbers of neurons and a log-sigmoid activation function. Finding: The correctness of the proposed stochastic computing scheme is observed through the overlapping of the results, best authentication values, and insignificant absolute error. Moreover, the statistical representations based on regression, transition state, error histogram, and function fitness also approve the reliability of the designed solver. Analysis: The numerical solutions of the maize streak epidemic model have been analyzed using different numbers of neurons to solve three distinct cases of the model. Novelty: The neuron analysis, along with the designed Bayesian regularization neural network, is proposed for the first time to solve the maize streak epidemic model with linear control.