Title: Predicting the enterprise tax risk using improved multilayer perceptive vector machine

Authors: Yi Liu

Addresses: School of Economics and Management, Changsha Normal University, Changsha, 410100, China

Abstract: With the comprehensive promotion of the business tax to value-added tax policy, the tax burden of enterprises is gradually reduced. Although office informatisation is progressing quickly, managing enterprise tax risk is still crucial. Multilayer perceptron can be combined with support vector machine to form multilayer perceptron vector machine. Therefore, the study uses the genetic algorithm to improve the multilayer perceptive vector machine, and on this basis, establishes the enterprise tax risk prediction model to improve the accuracy of tax risk prediction. According to experiment results, the CNN prediction model's accuracy in predicting economic risk, competitive risk, policy risk, and business risk is only 84.37%, while the accuracy of the improved algorithm was over 90% in all cases, with the accuracy of policy risk being as high as 95.87%. The results indicate that the improved algorithm can accurately predict the tax risks of enterprises, providing an effective method to guarantee the security of enterprise tax management.

Keywords: support vector machine; multilayer perceptron; genetic algorithm; tax risk.

DOI: 10.1504/IJWET.2023.133611

International Journal of Web Engineering and Technology, 2023 Vol.18 No.3, pp.185 - 201

Received: 29 Aug 2022
Received in revised form: 29 Jan 2023
Accepted: 02 Mar 2023
Published online: 25 Sep 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article