Modeling typhoid dynamics using recurrent neural networks with Bayesian regularization

The current research provides the numerical investigation of the epidemic typhoid model by using the stochastic artificial intelligence-based recurrent neural network. The typhoid system has five different categories: the susceptible, the carrier, the infected, the recovery and the bacterial population. The epidemic typhoid model is numerically handled by using the Bayesian regularization neural network. A Runge-Kutta procedure is used to get the dataset of the model, which is further trained through the Bayesian regularization. The separation of the dataset is performed through training 75 %, testing 15 % and validation 10 %. The precision of obtained performances of the epidemic typhoid model is perceived through the valuation of achieved and reference outcomes along with the absolute error calculated as 10^-06 to 10^-07. The attained actions of the model are specified to lessen the mean square error in input 10^-08-10^-10. To observe the efficiency of recurrent neural networks based on Bayesian regularization training, some tests like histogram error, state transitions and correlation indexes have been presented.