摘要
为了提高审计质量,建立了一种能够预测合并财务报表审计意见的改进MLP模型。此外,为了有效确定模型超参数,采用了dropout及自适应k-fold交叉验证进行训练,从而有效提升模型性能。通过仿真分析,提出的方法与传统MLP和RBF相比,性能有所提升,平均准确率达到92.54%;与CNN方法预测准确率提升不明显,然而本文方法模型复杂度较且和训练需求的资源更少。仿真结果表明所提方法能较好地预测合并财务报表审计意见。
In order to improve the audit quality,this paper provides an improved MLP model which can predict the audit opinion of consolidated financial statements.In addition,dropout and adaptive k-fold cross validation are used to improve the performance of the model.Through simulation analysis,compared with the traditional MLP and RBF,the performance of the proposed algorithm is better,is with an average accuracy of 92.54%;compared with CNN method,the prediction accuracy is not significantly improved,but the model complexity of the proposed method is lower and the resources required for training are less.The simulation results show that the proposed method can better predict the audit opinion of consolidated financial statements.
作者
邢凯华
XING Kaihua(School of Accounting, Xi’an Eurasia University, Xi’an 710065, China)
出处
《微型电脑应用》
2022年第2期49-52,共4页
Microcomputer Applications
基金
陕西省社科联2020年研究课题(20ZD195—110)。
关键词
审计质量
多层感知器
超参数
预测
audit quality
multi-layer perceptron
super parameter
prediction
作者简介
邢凯华(1979-),女,硕士,副教授,研究方向为管理会计、价值链。
