Fraud Detection in Credit Card Automated System using ML with AWS SageMaker

Abstract

Abstract: The article discusses the need for the widespread adoption of payment systems, which has been driven by advances in technology. However, the issue of fraud remains a major concern for financial institutions, as there is no one-size-fits-all approach to detecting and preventing fraudulent transactions. Machine learning has been identified as a potential solution to this problem, but it requires the development of a reliable automated system capable of handling large volumes of data in realtime. In the article, the author details the structure and setup of an automated fraud detection system for payment systems that relies on a web service hosted on the cloud. The deployment of this system is justified by utilizing Amazon Web Services as the platform, which includes Amazon Fraud Detector and Amazon A2I task type to authenticate and validate forecasts that are deemed high-risk. One instance of developing a system for detecting anomalies on Amazon DynamoDB streams is presented by utilizing AWS SageMaker, AWS Glue, and AWS Lambda. The software product aims to prevent and detect fraud in payment systems, with a rapid detection time and integration with various business institutions. The article also highlights the importance of developing a specific methodology for implementing the software product for fraud detection in payment systems.