Predicting and Preventing Fraud in Supply Chain Networks using Deep Neural Networks and Clustering Techniques

There is a lack of empirical evidence on the occurrence and causes of fraud in SCN, which poses a hidden but serious hazard to business effectiveness. There is a clear link between interorganizational fraud and relational risk in the SC, but because it is often covert, it goes undetected, which has serious operational and financial consequences. Using the viewpoints of agency theory, transaction cost economics, and the fraud triangle—frameworks commonly used in auditing and accounting—this research investigates five main causes of fraud in the supply chain. Organizations suffer a median revenue loss of 9% due to interorganizational fraud, according to survey data collected from 151 professionals in industrial supply chains. This study introduces a new technique to fixing the problems with traditional methods of fraud, which have high FPR and low TPR. Following extensive data processing, which included normalization and imputation of missing values, a CGOA was used to mimic natural grasshopper activity in order to optimize data features. The developed DCNNLSTM had a fraud detection accuracy rate of 94.60%. In addition to demonstrating the effectiveness of complex algorithmic models in improving fraud detection and prevention, our results stress the critical nature of addressing fraud in SCN.

Перевод пока недоступен