优化神经网络输入指标 提高财报舞弊识别效果

Improvement of Financial Fraud Recognition through Optimization of Neural Network Input Indicators

上市公司委托代理问题的存在,使得其投资者只能通过公司发布的财报信息评估公司的业绩,管理层从自身利益出发,有可能会利用财务舞弊手段,使得财报披露的盈余不再真实。通过使用BP神经网络,从舞弊发生的成因入手,运用合理的输入指标识别上市公司可能存在的财报舞弊,有效发现潜在的盈余管理及财务舞弊行为,从而保护投资者的利益。

Due to agency problems of listed companies,the only liable access to a company’s performance for investors is its financial statements.Financial fraud is frequently employed by the management out of interest concerns,which seriously distorts the company’s earnings disclosed in the financial statements.Starting from possible fraud causes,the BP neural network model applies reasonable input indicators to identify potential financial frauds of listed companies.Investors’interests can be effectively protected through identifying potential earnings management and financial frauds.