Real-Time Financial Fraud Graph Neural Network (GNN) Transaction Detector
Relational Graph Convolutional Network (RGCN) modeling banking entities as heterogeneous nodes to catch organized credit card fraud rings.
Project Overview
Constructs live heterogeneous graphs connecting users, bank accounts, device fingerprints, and IP addresses. Leverages PyTorch Geometric to learn relational embeddings, accurately detecting synthetic identity fraud and laundering networks that bypass traditional tabular rules.
Constructs live heterogeneous graphs connecting users, bank accounts, device fingerprints, and IP addresses. Leverages PyTorch Geometric to learn relational embeddings, accurately detecting synthetic identity fraud and laundering networks that bypass traditional tabular rules.
Relational Graph Convolutional Network (RGCN) modeling banking entities as heterogeneous nodes to catch organized credit card fraud rings.