Application Of Machine Learning To Parameter Estimation In The Propagation Dynamics Model Of Ebola

In recent years,with more and more studies on various infectious diseases,the work of parameter estimation is becoming more and more important.The methods mainly include least square method,MCMC method and machine learning,which can make up for the shortcomings of the other two methods.Along this line,this paper applies the machine learning method based on Gauss process to study the dynamics model of Ebola transmission.In the first chapter,we give the background of Ebola infectious diseases,the research background of several kinds of parameter estimation methods and the basic contents of machine learning methods.In the second chapter,we consider the dynamics of Ebola transmission with its own spread.Firstly,the model is discretized by backward Euler equation;Secondly,the variables in the Ebola propagation dynamics model are assumed to be a Gaussian process;Finally,by quasi Newton optimization method,we train the negative marginal likelihood function and get data fitting graph and parameter result.The parameters obtained in this chapter are close to those in other papers,which indicates that this method has applicability and provides a feasible method for parameter estimation in the future.In the third chapter,we consider the dynamics model of multidimensional Ebola transmission with cross diffusion.We use a machine learning method based on Gaussian process to estimate that the infection rate of susceptible individuals and the rate of spread of susceptible individuals and infected individuals.The results show that the machine learning method is suitable for multi-dimensional systems and the results are more accurate.In the forth chapter,we summarize and prospect the research contents of this paper.