Financial Fraud Identification Using Deep Learning Techniques

Abstract

The volume of internet transactions has rapidly increased in recent years. E-commerce and e-governance have both seen significant development in recent years. This has led to a rise in the number of persons adopting online payment options. The number of transactions that take place every day has increased exponentially as a result of this. The frequency of online scams has also increased as a result of the rise in online transactions. It is becoming more and more important to identify these fraudulent transactions as soon as possible in order to take the necessary measures and minimize any damages brought on by the fraud. In this study, ML models are proposed that might leverage previously known data and attempt to anticipate frauds using knowledge gained from the earlier data. Fraudulent actions often pose a danger to digital payment systems. Customers may avoid financial loss by having fraud transactions caught during money transfers. This article focuses on mobile-based money transfers for fraud detection. This research proposes a DL architecture for monitoring and identifying fraudulent activity. Deceptive transactions are found by implementing and using a RNN on a synthetic financial dataset produced by PaySim. The suggested technique has 99.87% accuracy, an F1-Score of 0.99for detecting illegitimate transactions.