Anomaly Detection and Data Analytics for Mobile Payment Activities via Scalable Graph Representation Learning


Introduction
Figure: Using a Graph Convolutional Network (GCN), one type of Graph Neural Networks (GNN), to detect anomalies from payment transaction graph.

With the rapid adoption of mobile payment services, the total value of payment transactions over Stored Value Facilities (SVF) in Hong Kong has exceeded HK$30 billion during Q1 2018. On one hand, such advancements bring new challenges to the SVF licensees and the regulators in safeguarding the proper uses of mobile payment services. On the other hand, unparalleled business opportunities will arise if one can harness the power to analyze/ leverage the uniquely rich datasets of mobile payment activities.

Towards these ends, we propose to design and implement a deep-neural-network-based graph learning system to support anomaly detection and Know-Your-Customer (KYC) analytics on large-scale online/ mobile payment networks. The system is designed to be scalable and customizable to handle diverse data inputs across different operating environments. By providing state-of-the-art performance in core graph-analytic tasks such as node classification and link prediction, our system will empower its users to perform scalable, accurate and cost-effective mining of payment activities of interest. The system will serve as an enabling technology for payment service providers, regulatory bodies and other stakeholders, to discover, analyze and monitor new/ unexpected usage patterns, abuses as well as illegal activities over large-scale online/ mobile payment networks in a practically feasible manner.


.: TOP :.
Last Updated on September 28 2022.
Copyright © 2022. All Rights Reserved. MobiTeC, The Chinese University of Hong Kong.
Disclaimer Privacy Statement