FEDERATED XGBOOST FOR ETHEREUM AML: PERFORMANCE EFFICIENCY TRADE OFF IN A MULTI EXCHANGE SETTING

Nazwa Auliarahman; Nur Widiyasono; Aradea

doi:10.24252/instek.v11i1.66853

Authors

Nazwa Auliarahman Universitas Siliwangi
Nur Widiyasono Universitas Siliwangi
Aradea Universitas Siliwangi

DOI:

https://doi.org/10.24252/instek.v11i1.66853

Keywords:

Anti Money Laundering, Blockchain, Centralized, Federated

Abstract

Penelitian ini mengevaluasi efektivitas federated learning dalam mendeteksi alamat ilegal pada blockchain Ethereum untuk Anti Money Laundering (AML). Studi ini membandingkan XGBoost centralized dan federated dalam simulasi multi exchange, data dibagi secara horizontal menjadi 3 subset yang merepresentasikan entitas bursa independen. Untuk mengisolasi pengaruh pelatihan terdistribusi, pembagian data dilakukan secara homogen (IID), sehingga analisis difokuskan pada perbedaan mekanisme pembelajaran tanpa dipengaruhi heterogenitas data. Hasil menunjukkan bahwa model federated mencapai performa yang kompetitif dengan PR AUC 0,9962 dan akurasi 97,11%, dibandingkan model terpusat dengan PR AUC 0,9975 dan akurasi 97,75%. Namun, performa tersebut disertai peningkatan durasi pelatihan 4,01 detik dibandingkan 3,39 detik, yang disebabkan oleh beban komunikasi selama proses pembaruan model. Temuan ini menegaskan adanya trade off antara kinerja dan efisiensi dalam penerapan federated learning. Meskipun mampu mendekati performa pembelajaran centralized, pendekatan ini menimbulkan biaya operasional tambahan. Studi ini merupakan evaluasi dasar dalam kondisi terkontrol dan belum merepresentasikan skenario dunia nyata, sehingga diperlukan penelitian lanjutan pada data non IID dan skala sistem yang lebih besar.

Downloads

Download data is not yet available.

References

[1] The Financial Action Task Force (FATF), “Risk Based Approach Guidance for the Real Estate Sector,” Fatf, no. July, 2022, [Online]. Available: https://www.fatf-gafi.org/en/publications/Fatfrecommendations/Guidance-rba-real-estate-sector.html

[2] L. Pahuja and A. Kamal, “EnLEFD-DM: Ensemble Learning based Ethereum Fraud Detection using CRISP-DM framework,” Expert Syst., vol. 40, no. 9, pp. 1–18, 2023, doi: 10.1111/exsy.13379.

[3] Z. Gu and O. Dib, “Enhancing fraud detection in the Ethereum blockchain using ensemble learning,” PeerJ Comput. Sci., vol. 11, 2025, doi: 10.7717/PEERJ-CS.2716.

[4] C. Zhang, Y. Xie, H. Bai, B. Yu, W. Li, and Y. Gao, “A survey on federated learning,” Knowl. Based. Syst., vol. 216, p. 106775, 2021, doi: 10.1016/j.knosys.2021.106775.

[5] B. Guembe, A. Azeta, V. Osamor, and R. Ekpo, “A Federated Machine Learning Approaches For Anti-Money Laundering Detection,” SSRN Electronic Journal, vol. 1, pp. 1–13, 2022, doi: 10.2139/ssrn.4669561.

[6] C. Ma, X. Qiu, D. Beutel, and N. Lane, “Gradient-less Federated Gradient Boosting Tree with Learnable Learning Rates,” pp. 56–63, 2023, doi: 10.1145/3578356.3592579.

[7] A. A. Ahmed and O. O. Alabi, “Secure and Scalable Blockchain-Based Federated Learning for Cryptocurrency Fraud Detection: A Systematic Review,” IEEE Access, vol. 12, no. August, pp. 102219–102241, 2024, doi: 10.1109/ACCESS.2024.3429205.

[8] W. Lindskog and C. Prehofer, “A Federated Learning Benchmark on Tabular Data: Comparing Tree-Based Models and Neural Networks,” 2023 8th International Conference on Fog and Mobile Edge Computing, FMEC 2023, pp. 239–246, 2023, doi: 10.1109/FMEC59375.2023.10305887.

[9] T. Fan, W. Chen, G. Ma, Y. Kang, L. Fan, and Q. Yang, “SecureBoost+: Large Scale and High-Performance Vertical Federated Gradient Boosting Decision Tree,” Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14647 LNAI, pp. 237–249, 2024, doi: 10.1007/978-981-97-2259-4_18.

[10] C. C. Wang, P. H. Kuo, and G. Y. Chen, “Machine Learning Prediction of Turning Precision Using Optimized XGBoost Model,” Applied Sciences (Switzerland), vol. 12, no. 15, 2022, doi: 10.3390/app12157739.

[11] S. Garst, J. Dekker, and M. Reinders, “A comprehensive experimental comparison between federated and centralized learning,” Database, vol. 2025, no. 00, pp. 1–10, 2025, doi: 10.1093/database/baaf016.

[12] J. Liu et al., “Graph Embedding-Based Money Laundering Detection for Ethereum,” Electronics (Switzerland), vol. 12, no. 14, pp. 1–14, 2023, doi: 10.3390/electronics12143180.

[13] C. Y. Dan Lin, Jiajing Wu, Yunmei Yu, Qishuang Fu, Zibin Zheng, “DenseFlow: Spotting Cryptocurrency Money Laundering in Ethereum Transaction Graphs,” Proceedings of the ACM Web Conference 2024, 2024, [Online]. Available: https://github.com/DenseFlow.

[14] K. Song et al., Identifying Money Laundering Subgraphs on the Blockchain, vol. 1, no. 1. Association for Computing Machinery, 2024. doi: 10.1145/3677052.3698635.

[15] K. Cheng et al., “SecureBoost: A Lossless Federated Learning Framework,” IEEE Intell. Syst., vol. 36, no. 6, pp. 87–98, 2021, doi: 10.1109/MIS.2021.3082561.