THE RESEARCH AND APPLICATION OF TAX RISK IDENTIFICATION BASED ON PSO-BP ALGORITHM

Tax risk management is a new business field of tax management, which has become the core task of professional revenue management. As the traditional tax payment model has shortcoming of inefficiency and high cost, it is necessary to create a stable and orderly management environment to identify tax risks for different taxpayers quickly and efficiently. In this paper, the theory of machine learning is introduced in the practice of tax risk management, and the model of tax risk identification based on PSO-BP (Particle Swarm Optimization-Back Propagation) algorithm is established. 1350 tax households are randomly selected in this paper. Among these tax households, 1000 of them are training data, and the rest of them are testing data. Compared with standard BP algorithm, the experimental results show that the optimized PSO-BP algorithm can identify the possible tax risk category effectively and accurately, which realize great improvement of the tax inspection.