The Artificial Neural Network Is In The Individual Tax Of Applied Research

Artificial Neural Network (ANN) is the simulation of brain information processor mechanism. There is a potential worthiness of application in intelligent systems, because of ANN's abilities of self-organizing, self- learning, self-adapting, nonlinear approximating, large-scale parallel processing and so on.The individual economy is the main part of the local revenue. However, individual constant tax has much subjectivity, and it to be the difficulty in the work of individual revenue. Coefficient constant tax enhances the fairness of the revenue, but the model can not real-time and dynamically compute the factual tax. What's more, there is a limitation that the precision is not high in computing the tax with the software of coefficient constant tax.According to the question, we propose a solution based on Artificial Neural Network technology to forecast individual constant tax. We carry out a lot of experimentation with the original data provided by ZhuZhou Local Taxation Bureau, then analyze, design and implement the Intelligent Constant Tax System, which provides a new idea for the work of individual constant tax.In this paper, we mainly discuss the application of ANN to individual constant tax. Firstly, we introduce Artificial Neural Network technology and several primary network learning algorithms, mainly about BP algorithm and Cascade-Correlation algorithm. And then proposes BP and Cascade-Correlation leaning algorithm which applied in the individual constant tax. From the experiments, we found that The BP algorithm is very effective, which indicated that it is feasible that the technology of neural network be used in individual constant tax and can resolve many problems in a certain extent, such as multi-factor, uncertainty, non-linear and so on. According to the question that the structure of BP neural network was very difficult to be ascertained and the precision of its' forecast result made it can not have the ability to dynamic satisfy the necessary of constant tax. Cascade-Correlation algorithm resolves many problems that BP algorithm can not solve, and it enhances the self-adaptive of network structure and forecasting precision of constant tax. Meanwhile, we design and exploit the intelligent constant tax of the system, which based on Matlab Web Server, and contact it with the Local Taxation software, which will afford a flat roof for constant tax work.