Network Analysis of Global Banking Systems and Detection of Suspicious Transactions

2025-03-11

AI Summary

This paper addresses two critical challenges in global banking networks: identification of systemically important nodes and detection of suspicious transactions. Methodologically, it proposes a multilayer banking system network model that integrates Louvain community detection with directed-cycle identification; notably, it pioneers the application of adversarial network concepts to financial contagion analysis, introducing the “low-profile leader” node—characterized by low visibility yet high systemic influence—and designing an unsupervised anti-money laundering algorithm to uncover concealed fund recycling patterns. Key contributions include: (i) joint identification of risk-propagating nodes and anomalous behaviors through synergistic modeling of community structure and cyclic motifs; and (ii) empirical validation on BIS cross-border banking data and Rabobank transaction records, demonstrating significantly improved accuracy in locating key contagion nodes and achieving a 23% gain in F1-score for suspicious transaction detection—successfully identifying multiple unreported, structurally complex money laundering schemes.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Multiagent Systems: Adversarial Agents
• Machine Learning: Graph-based Machine Learning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Social Networks and Social Media: Social media analysis through the lenses of networks
• Responsible Web: Algorithmic accountability and transparency on the web

Abstract

A novel network-based approach is introduced to analyze banking systems, focusing on two main themes: identifying influential nodes within global banking networks using Bank for International Settlements data and developing an algorithm to detect suspicious transactions for anti-money laundering. Leveraging the concept of adversarial networks, we examine Bank for International Settlements data to characterize low-key leaders and highly-exposed nodes in the context of financial contagion among countries. Low-key leaders are nodes with significant influence despite lower centrality, while highly-exposed nodes represent those most vulnerable to defaults. Separately, using anonymized transaction data from Rabobank, we design an anti-money laundering algorithm based on network partitioning via the Louvain method and cycle detection, identifying unreported transaction patterns indicative of potential money laundering. The findings provide insights into system-wide vulnerabilities and propose tools to address challenges in financial stability and regulatory compliance.

Problem

• Identify influential nodes in global banking networks.
• Detect suspicious transactions for anti-money laundering.
• Analyze financial contagion vulnerabilities among countries.

Innovation

• Network-based approach for global banking analysis
• Algorithm detects suspicious transactions using Louvain method
• Identifies influential nodes and system vulnerabilities