基于卷积神经网络的企业会计财务核算数据误差校正方法研究

Research on Error Correction Method of Enterprise Accounting Data Based on Convolutional Neural Network

面对企业会计财务核算过程中存在的数据误差问题,传统误差校正方法在网络噪声环境下,容易出现数据丢失情况,影响财务核算数据的误差校正效果,校正后的数据可靠性需要进一步提高.针对此问题,提出基于卷积神经网络的企业会计财务核算数据误差校正方法.首先,分解原始财务核算数据集合,确定财务核算数据是否存在误差;其次,以卷积神经网络为基础,设计财务核算数据分类模型,将含有误差的数据作为该模型的输入,利用模型内部卷积层和池化层,确定核算数据之间的关系;最后,在此基础上,预测核算数据下一状态变化,引入外控制量并求解,根据控制量的变化校正数据误差.实验结果表明:在网络噪声环境下,基于卷积神经网络的误差校正方法数据补偿效果好,数据方差小,数据辍出率在5%以下.

In the face of the data error problem in the process of enterprise financial accounting,the traditional error correction method is prone to data loss in the network noise environment,which affects the error correction effect of financial accounting data,and the data reliability of the method to be further improved.Aiming at this problem,an error correction method for enterprise financial accounting data based on convolutional neural network is proposed.The original data set is decomposed to determine whether there are errors in the financial accounting data.Based on the convolution neural network,the classification model of financial accounting data is designed.The data containing errors are used as the input of the model.The relationship between the accounting data is determined by using the convolution layer and pooling layer within the model.On this basis,the next state change of the accounting data is predicted,and the external control is introduced and solved.The data error is corrected according to the change of the control.The experimental results show that the error correction method based on convolutional neural network has good data compensation effect,small data variance and less than 5%data dropout rate in the network noise environment.