FRAUD DETECTION OF AUDIT DATA BASED ON WEIGHTED KNN AND COST-SENSITIVE MULTI-BRANCH DEEP NEURAL NETWORK
In the face of increasing audit objectivity and increasing audit tasks,it is imperative to improve the efficiency and quality of audit.In this paper,the financial voucher data of an enterprise in the power industry is selected as the research object.Aimed at the problems of large number of financial documents,diverse data types and serious imbalance of positive and negative data samples,a new algorithm based on weighted KNN and cost-sensitive multi-branch deep neural network is proposed.It could effectively reduce the scope of verification,and the obtained financial documents with audit doubtful points covered as much audit problems as possible in order to help auditors improve their work efficiency.Through comparative experiments,it is verified that the proposed algorithm can effectively find audit doubts and cover audit problems,and the results are better than other existing methods.
Intelligent auditMachine learningArtificial intelligenceFraud detectionCost-sensitiveMulti-branch deep neural network
万方数据
维普
