Integrating Sequential Information and Graph Structures for Anti-Money Laundering Anomaly Detection

Yin-Ju Wu, Gavin Tseng, Berlin Chen


Abstract
反洗錢(Anti-Money Laundering, AML)是金融科技領域的重要研究課題,其目標在於識別潛在的可疑帳戶與交易。然而隨著跨境支付與新型態交易的興起,洗錢行為往往具有高度隱匿性與複雜的網路結構,傳統規則式方法在偵測效能與泛化能力上皆表現不足。近年來,雖然有研究嘗試將機器學習或深度學習方法應用於 AML,但仍存在許多挑戰。為了解決這些問題,本研究提出一個基於序列圖融合的 AML 帳戶風險預測框架。該方法的核心在於同時建模帳戶的個體時序行為與其在交易網路中的結構特徵。首先,將每個帳戶的交易歷史分解為入邊和出邊序列,使用雙分支GRU架構分別編碼,捕捉帳戶的時序交易模式,接著使用雙向注意力圖卷積層,通過差異感知的消息傳遞機制同時處理正向和反向鄰居關係,學習帳戶間的行為差異,並通過注意力機制自適應融合節點自身特徵與雙向鄰居聚合特徵。此外,針對 AML 資料集的極度不平衡特性,引入類別重加權與平衡採樣策略。我們在公開的反洗錢資料集上驗證所提方法,實驗結果顯示該框架在極度不平衡的情境下能取得穩定的 F1 表現,相較於傳統基線方法具有顯著優勢。
Anthology ID:
2025.rocling-main.34
Volume:
Proceedings of the 37th Conference on Computational Linguistics and Speech Processing (ROCLING 2025)
Month:
November
Year:
2025
Address:
National Taiwan University, Taipei City, Taiwan
Editors:
Kai-Wei Chang, Ke-Han Lu, Chih-Kai Yang, Zhi-Rui Tam, Wen-Yu Chang, Chung-Che Wang
Venue:
ROCLING
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
320–325
Language:
URL:
https://aclanthology.org/2025.rocling-main.34/
DOI:
Bibkey:
Cite (ACL):
Yin-Ju Wu, Gavin Tseng, and Berlin Chen. 2025. Integrating Sequential Information and Graph Structures for Anti-Money Laundering Anomaly Detection. In Proceedings of the 37th Conference on Computational Linguistics and Speech Processing (ROCLING 2025), pages 320–325, National Taiwan University, Taipei City, Taiwan. Association for Computational Linguistics.
Cite (Informal):
Integrating Sequential Information and Graph Structures for Anti-Money Laundering Anomaly Detection (Wu et al., ROCLING 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.rocling-main.34.pdf