WeirdFlows: Anomaly Detection in Financial Transaction Flows

2025-03-20

AI Summary

This paper addresses the challenge of interpretable anomaly detection for money laundering and sanctions evasion in high-dynamic, unlabeled anti-financial crime (AFC) environments. It tackles three key challenges: weak model interpretability, severe scarcity of labeled data, and time-varying, complex fraud patterns. To this end, we propose the first top-down, pattern-agnostic, fully unsupervised search framework—incorporating dynamic directed transaction graph modeling, subgraph-level anomaly scoring, temporal sparse pattern mining, and root-cause path attribution for explanation. Evaluated on 80 million real-world cross-border transaction records, our method successfully identified novel sanctions circumvention behaviors emerging after the Russia-Ukraine conflict, with accuracy and interpretability independently validated by domain experts from major banks. Our core contribution is a new paradigm for unsupervised, interpretable, and temporally adaptive anomaly detection—requiring neither predefined templates nor labeled training data.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Unsupervised & Self-Supervised Learning
• Reasoning under Uncertainty: Causality

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Bridging structured and unstructured data
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

In recent years, the digitization and automation of anti-financial crime (AFC) investigative processes have faced significant challenges, particularly the need for interpretability of AI model results and the lack of labeled data for training. Network analysis has emerged as a valuable approach in this context. In this paper, we present WeirdFlows, a top-down search pipeline for detecting potentially fraudulent transactions and non-compliant agents. In a transaction network, fraud attempts are often based on complex transaction patterns that change over time to avoid detection. The WeirdFlows pipeline requires neither an a priori set of patterns nor a training set. In addition, by providing elements to explain the anomalies found, it facilitates and supports the work of an AFC analyst. We evaluate WeirdFlows on a dataset from Intesa Sanpaolo (ISP) bank, comprising 80 million cross-country transactions over 15 months, benchmarking our implementation of the algorithm. The results, corroborated by ISP AFC experts, highlight its effectiveness in identifying suspicious transactions and actors, particularly in the context of the economic sanctions imposed in the EU after February 2022. This demonstrates extit{WeirdFlows}' capability to handle large datasets, detect complex transaction patterns, and provide the necessary interpretability for formal AFC investigations.

Problem

• Detects fraudulent transactions in financial networks.
• Identifies non-compliant agents without predefined patterns.
• Provides interpretable results for anti-financial crime investigations.

Innovation

• Top-down search pipeline for fraud detection
• No need for predefined patterns or training data
• Provides interpretable anomaly explanations for analysts