DynBERG: Dynamic BERT-based Graph neural network for financial fraud detection

2025-10-28

AI Summary

Financial fraud detection faces challenges in modeling dynamic, directed transaction graphs: existing Graph-BERT methods are designed for static, undirected graphs and cannot capture directional fund flows or temporal evolution. To address this, we propose DynBERG—the first Graph-BERT extension for dynamic directed graph-based fraud detection—incorporating GRUs to explicitly model multi-step temporal node representation evolution and adapting edge aggregation to preserve directionality. Evaluated on the Elliptic dataset, DynBERG significantly outperforms EvolveGCN and GCN, maintaining strong robustness before and after market shocks. Ablation studies confirm the critical role of GRUs in temporal modeling. This work marks the first successful adaptation of Graph-BERT to dynamic cryptocurrency transaction scenarios (e.g., Bitcoin), establishing a scalable, temporally aware graph learning paradigm for decentralized finance risk management.

Technology Category

• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Financial fraud detection is critical for maintaining the integrity of financial systems, particularly in decentralised environments such as cryptocurrency networks. Although Graph Convolutional Networks (GCNs) are widely used for financial fraud detection, graph Transformer models such as Graph-BERT are gaining prominence due to their Transformer-based architecture, which mitigates issues such as over-smoothing. Graph-BERT is designed for static graphs and primarily evaluated on citation networks with undirected edges. However, financial transaction networks are inherently dynamic, with evolving structures and directed edges representing the flow of money. To address these challenges, we introduce DynBERG, a novel architecture that integrates Graph-BERT with a Gated Recurrent Unit (GRU) layer to capture temporal evolution over multiple time steps. Additionally, we modify the underlying algorithm to support directed edges, making DynBERG well-suited for dynamic financial transaction analysis. We evaluate our model on the Elliptic dataset, which includes Bitcoin transactions, including all transactions during a major cryptocurrency market event, the Dark Market Shutdown. By assessing DynBERG's resilience before and after this event, we analyse its ability to adapt to significant market shifts that impact transaction behaviours. Our model is benchmarked against state-of-the-art dynamic graph classification approaches, such as EvolveGCN and GCN, demonstrating superior performance, outperforming EvolveGCN before the market shutdown and surpassing GCN after the event. Additionally, an ablation study highlights the critical role of incorporating a time-series deep learning component, showcasing the effectiveness of GRU in modelling the temporal dynamics of financial transactions.

Problem

• Detecting financial fraud in dynamic cryptocurrency transaction networks
• Addressing limitations of static graph models for evolving financial structures
• Capturing temporal dynamics and directed edges in fraud detection

Innovation

• Dynamic BERT-based Graph neural network for fraud detection
• Integrates Graph-BERT with GRU to capture temporal evolution
• Modifies algorithm to support directed edges in transactions