摘要
为提高上市公司财务舞弊识别模型判断的准确度,文章以1998—2009年中国证监会网站上公开披露的财务舞弊的56家上市公司的75个舞弊年度为研究对象,并选取了与舞弊公司同行业、同年度的75个非舞弊上市公司年度作为控制样本,运用量子粒子群算法改进的BP神经网络建立财务舞弊的识别模型。研究结果表明,采用量子粒子群算法与BP神经网络结合的方法建立的财务舞弊识别模型判断的准确度较标准BP神经网络判断的准确度有了较大提高。这对于防范上市公司财务舞弊、提高监管效率、降低投资者的损失等方面具有一定的积极意义。
To improve the accuracy of listed companies' identification of fraudulent financial reporting,this paper selects 75 fraud years of 56 listed companies were disclosed public by the China Securities Regulatory Commission from 1998 to 2009 for research units, Meanwhile,the paper selects 75 non-fraud listed companies for controlling with the same industry and the same year of fraud listed companies. We use the method of quantum particle swarm algorithm and back propagation neural network to establish the model of recogning financial fraud. The results show that the model of fraudulent financial reporting identification with the quantum particle swarm algorithm and back propagation neural network method has improved in accuracy. It has certain positive meanings for preventing listed companies' financial fraud, enhancing regulatory efficiency ,and reducing the loss of investors and so on.
出处
《科技与管理》
2011年第1期108-111,116,共5页
Science-Technology and Management
基金
教育部人文社会科学规划项目(10XJA630182)
关键词
财务舞弊
审计报告
BP神经网络
量子粒子群算法
fraudulent financial reporting
auditing report
back propagation network
quantum particle swarmoptimization algorithm