Artificial Neural Networks In The Tax Assessment

The CTAIS has been widely applied by all levels of tax administration around China.Though the good results are obtained,yet the ratepaying evaluations of CTAIS are still adopting manual evaluation,not automatic evaluation.Whether the ratepaying evaluation can be done automatically will have direct influence on the intelligent level of the information system on tax collection and management as well as on the accuracy of information-based management.It is a very important part of the information system on tax collection and management.This paper focuses on the ratepaying evaluation of the information system on tax collection and management of Shixi Road Market of National Tax Bureau of Yunyan District,Guiyang City, Guizhou Province.It carefully analyzes the problems that the ratepaying evaluation has based on the binary linear regression algorithm,and probes into the advantages of artificial neural network algorithm compared with the multiple linear regression algorithm.And then it goes deeply into Multilayer Perceptrons of the artificial neural network,BP learning algorithm and some improved algorithm of BP learning algorithm,which are followed by the research of the fulfillment of MATLAB of BP network.The researches also cover the selecting of different learning methods on one network model,the confirmation of the network structure,and the adoption of the parameter like the study efficiency and the initial value.Afterwards,it introduces the design and the fulfillment of ratepaying evaluation based on BP neural network.This paper finally comes to an conclusion and proposes that further research of the application of artificial neural network on ratepaying evaluation is needed.The paper's biggest feature is to apply the neural network technique to the ratepaying evaluation to realize its automation,which brings about certain pragmatic value..