深度卷积神经网络对领导干部自然资源资产离任审计疑点提取的应用  

Research on the Extraction and Analysis of Doubtful Points in the Audit of Natural Resources Assets Leaving Office of Leading Cadres Based on Deep Convolution Neural Network

作　　者：张丽达 谢宸湘 ZHANG Li-da;XIE Chen-xiang(School of Business,Xi?an University of Finance and Economics,Xi?an 710100,China;Shaanxi Aerospace Technology Applied Research Institute Co.,Ltd.,Xi'an 710100,China)

机构地区：[1]西安财经大学商学院,陕西西安710100 [2]陕西航天技术应用研究院有限公司,陕西西安710100

出　　处：《统计与信息论坛》2021年第8期75-83,共9页Journal of Statistics and Information

基　　金：国家社会科学基金青年项目“基于演化仿真的社会化媒体舆论信息与审计师行为研究”(18CJK003)。

摘　　要：针对领导干部自然资源资产离任审计工作中,通过人工识别提取自然资源资产实物量的变化来反映审计疑点的效率低下及主观意识性强的问题,采用DeepLab深度卷积神经网络算法基于卫星影像提取地物信息,从审计时间区间内的卫星影像上提取数据,并结合自然资源管理的规划图斑进行比对分析,采用反举证法进行疑点批量化、自动化处理,从而快速发现审计疑点。通过对研究区域林地资源的应用实践证明深度卷积神经网络模型在提取审计疑点上效果良好,与人工提取结果一致。因此,充分利用该模型可以替代人工提取审计疑点,很大程度上优化了审计疑点提取中人力物力的投入,从而更好地利用实验人工智能,提高审计效率。Big data technology is more and more widely used in audit.In view of the audit of the new leading cadres’natural resources assets audit project,the technical application of the extraction of audit doubts is also a difficult problem to be solved.This paper aims at the problems of the change of the physical quantity of natural resources assets in the audit work of leading cadres’natural resources assets leaving office,such as the low efficiency and strong subjective consciousness of extracting audit doubts,based on the convolution neural network algorithm,extracting data from the satellite images in the audit time interval,combining with the distribution and planning data of natural resources,using the counter proof method to carry out the batch of doubts.Put forward technology of automation.The experimental results show that the effect is good,and highly fit with the artificial extraction,which greatly optimizes the input of human and material resources for the audit query extraction,and makes better use of the experimental artificial intelligence work.

关 键 词：大数据技术 深度卷积神经网络 自然资源资产审计 

分 类 号：F234[经济管理—会计学]