Title: Enterprise financial anomaly data detection method based on improved support vector machine

Authors: Hao Wang; Huan Wang

Addresses: College of Financial Management, Henan Polytechnic, Zhengzhou, 450018, China ' College of Automobile and Transportation, Henan Polytechnic, Zhengzhou, Henan, 450018, China

Abstract: In the process of detecting abnormal financial data in enterprises, due to the suboptimal detection results, there are problems with low fitting and large detection errors in the detection algorithm. Therefore, an improved support vector machine-based method for detecting abnormal financial data in enterprises is proposed. By analysing the financial abnormal data of enterprises, the distribution of financial abnormal data is determined. The text CNN neural network model is used to extract initial abnormal data features, and the least squares method is used to adjust the nonlinear features of abnormal data to achieve feature extraction of financial abnormal data of enterprises. The support vector machine algorithm is optimised by differential evolution to determine the optimal classification population of enterprise financial abnormal data features for global optimisation, so as to realise the detection of enterprise financial abnormal data. The test results show that the determinability coefficient of this method is always close to 1, and the error rate of anomaly data detection is less than 0.5%, indicating that the method in this paper has a good fitting degree, low detection error, and is feasible.

Keywords: improving support vector machines; point anomaly; context exception; collection anomaly; least squares method; differential evolution algorithm.

DOI: 10.1504/IJRIS.2025.149721

International Journal of Reasoning-based Intelligent Systems, 2025 Vol.17 No.6, pp.364 - 372

Received: 05 Jun 2023
Accepted: 09 Aug 2023
Published online: 11 Nov 2025 *

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