基于GAN的高企认定申报数据异常检测模型

Abnormal Detection Model of High Enterprise Identification and Declaration Data Based on GAN

高新技术企业认定需要审核的数据繁多,企业在申报过程中经常出现错填、漏填、数据出错、数据不完整等数据异常问题,影响高新技术企业的正常评定程序。通过对广东省高新技术企业认定系统已有数据的研究分析,提出一种基于门控循环神经网络和生成对抗网络的高新技术企业认定申报数据异常检测模型。基于生成对抗网络(GAN)的申报数据异常检测模型在通过生成网络G学习正常样本的分布,使用判别网络D来判别申报数据是不是“真实的”,从而实现数据异常检测。在高新技术企业认定事项管理数据集上进行了实验,实验结果证明了本文提出的模型优于其他模型。

There are a lot of data that need to be reviewed for the identification of high-tech enterprises. In the process of declaration, enterprises often have abnormal data problems such as misfiling, missing filling, data error and incomplete data, which affect the normal evaluation procedure of high-tech enterprises. Based on the research and analysis of the existing data of the identification system of high-tech enterprises in Guangdong Province, this paper proposes an anomaly detection model based on the gated recurrent neural network and the generative adversarial network. The reported data anomaly detection model based on Generative Adversarial Network (GAN) learns the distribution of normal samples by generating network G, and uses discriminant network D to determine whether the reported data is “real”, so as to realize data anomaly detection. The experiment is carried out on the data set of the identification of high-tech enterprises, and the experimental results prove that the model proposed in this paper is superior to other models.