Federated Graph Learning for Private Inter-Bank Anti-Money Laundering
Privacy-preserving graph neural network sharing transaction relationship weights across multiple institutions without sharing customer account data.
Project Overview
Enables multiple financial institutions to collaboratively train an anti-money laundering (AML) model. Uses Subgraph Federated Learning and secure aggregation protocols to detect cross-bank transaction cycles without exposing sensitive customer identity records.
Enables multiple financial institutions to collaboratively train an anti-money laundering (AML) model. Uses Subgraph Federated Learning and secure aggregation protocols to detect cross-bank transaction cycles without exposing sensitive customer identity records.
Privacy-preserving graph neural network sharing transaction relationship weights across multiple institutions without sharing customer account data.