Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost

Joao Crisostomo; Victor Lobo; Fernando Bacao

doi:10.1007/978-3-031-45155-3_23

Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost

Joao Crisostomo, Victor Lobo, Fernando Bacao

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The illicit activity in Blockchain reached an all-time high in 2021. In this work, we combined two machine learning techniques, Autoencoder (AE) and Extreme Gradient Boosting (XGBoost), to improve the performance of predicting illicit activity at the account level. The choice of autoencoding technique allows us to be able to detect new MOs (modus operandi) from fraudsters. With an Autoencoder trained only with healthy accounts, we are not misleading the model to focus on specific MOs as it happens with tree-based models. This allows us to introduce a dimension that could capture future fraudulent behaviours. Furthermore, the dataset was generated considering the real applicability of the model, i.e. it mimics what can realistically be obtained in a practical situation. With this approach, we were able to improve the state-of-the-art performance. In our test set the precision-recall AUC (area under the curve) of our final model increased by 12,07% when compared with our baseline.

Original language	English
Title of host publication	Blockchain and Applications, 5th International Congress
Subtitle of host publication	BLOCKCHAIN 2023
Editors	José Manuel Machado, Paulo Vieira, António Abelha, Luigi Vigneri, Javier Prieto, Hugo Peixoto, David Arroyo
Place of Publication	Gewerbestrasse
Publisher	Springer, Cham
Pages	224-233
Number of pages	10
ISBN (Electronic)	978-3-031-45155-3
ISBN (Print)	978-3-031-45154-6
DOIs	https://doi.org/10.1007/978-3-031-45155-3_23
Publication status	Published - 14 Nov 2023
Event	5th International Congress on Blockchain and Applications - Hotel de Guimarães, Guimarães, Portugal Duration: 12 Jul 2023 → 14 Jul 2023 Conference number: 5 https://edition2023.blockchain-congress.net/

Publication series

Name	Lecture Notes in Networks and Systems
Publisher	Springer Cham
Volume	778
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Conference

Conference	5th International Congress on Blockchain and Applications
Abbreviated title	BLOCKCHAIN'23
Country/Territory	Portugal
City	Guimarães
Period	12/07/23 → 14/07/23
Internet address	https://edition2023.blockchain-congress.net/

Keywords

Anomaly detection
Ethereum
Blockchain
Autoencoder
XGBoost
Fraud Detection
Malicious Accounts
Machine Learning

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-031-45155-3_23

Embargoed Document

Detecting fraudulent wallets in Ethereum blockchain combining supervised and unsupervised techniques - using Autoencoders and XGboost
Accepted author manuscript, 749 KB
Licence: Unspecified
Embargo ends: 15/11/24

Cite this

Crisostomo, J., Lobo, V., & Bacao, F. (2023). Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost. In J. M. Machado, P. Vieira, A. Abelha, L. Vigneri, J. Prieto, H. Peixoto, & D. Arroyo (Eds.), Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023 (pp. 224-233). (Lecture Notes in Networks and Systems; Vol. 778). Springer, Cham. https://doi.org/10.1007/978-3-031-45155-3_23

Crisostomo, Joao ; Lobo, Victor ; Bacao, Fernando. / Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques : Using Autoencoders and XGboost. Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023. editor / José Manuel Machado ; Paulo Vieira ; António Abelha ; Luigi Vigneri ; Javier Prieto ; Hugo Peixoto ; David Arroyo. Gewerbestrasse : Springer, Cham, 2023. pp. 224-233 (Lecture Notes in Networks and Systems).

@inproceedings{cf4c3895c0e84140b0faf41b80f58837,

title = "Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost",

abstract = "The illicit activity in Blockchain reached an all-time high in 2021. In this work, we combined two machine learning techniques, Autoencoder (AE) and Extreme Gradient Boosting (XGBoost), to improve the performance of predicting illicit activity at the account level. The choice of autoencoding technique allows us to be able to detect new MOs (modus operandi) from fraudsters. With an Autoencoder trained only with healthy accounts, we are not misleading the model to focus on specific MOs as it happens with tree-based models. This allows us to introduce a dimension that could capture future fraudulent behaviours. Furthermore, the dataset was generated considering the real applicability of the model, i.e. it mimics what can realistically be obtained in a practical situation. With this approach, we were able to improve the state-of-the-art performance. In our test set the precision-recall AUC (area under the curve) of our final model increased by 12,07% when compared with our baseline.",

keywords = "Anomaly detection, Ethereum, Blockchain, Autoencoder, XGBoost, Fraud Detection, Malicious Accounts, Machine Learning",

author = "Joao Crisostomo and Victor Lobo and Fernando Bacao",

note = "Cris{\'o}stomo, J., Lobo, V., & Ba{\c c}{\~a}o, F. (2023). Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost. In J. M. Machado, P. Vieira, A. Abelha, L. Vigneri, J. Prieto, H. Peixoto, & D. Arroyo (Eds.), Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023 (pp. 224-233). (Lecture Notes in Networks and Systems; Vol. 778). Springer, Cham. https://doi.org/10.1007/978-3-031-45155-3_23; 5th International Congress on Blockchain and Applications, BLOCKCHAIN'23 ; Conference date: 12-07-2023 Through 14-07-2023",

year = "2023",

month = nov,

day = "14",

doi = "10.1007/978-3-031-45155-3_23",

language = "English",

isbn = " 978-3-031-45154-6",

series = "Lecture Notes in Networks and Systems",

publisher = "Springer, Cham",

pages = "224--233",

editor = "Machado, {Jos{\'e} Manuel} and Paulo Vieira and Ant{\'o}nio Abelha and Luigi Vigneri and Javier Prieto and Hugo Peixoto and David Arroyo",

booktitle = "Blockchain and Applications, 5th International Congress",

url = "https://edition2023.blockchain-congress.net/",

}

Crisostomo, J , Lobo, V & Bacao, F 2023, Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost. in JM Machado, P Vieira, A Abelha, L Vigneri, J Prieto, H Peixoto & D Arroyo (eds), Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023. Lecture Notes in Networks and Systems, vol. 778, Springer, Cham, Gewerbestrasse, pp. 224-233, 5th International Congress on Blockchain and Applications, Guimarães, Portugal, 12/07/23. https://doi.org/10.1007/978-3-031-45155-3_23

Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost. / Crisostomo, Joao ; Lobo, Victor ; Bacao, Fernando.
Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023. ed. / José Manuel Machado; Paulo Vieira; António Abelha; Luigi Vigneri; Javier Prieto; Hugo Peixoto; David Arroyo. Gewerbestrasse: Springer, Cham, 2023. p. 224-233 (Lecture Notes in Networks and Systems; Vol. 778).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques

T2 - 5th International Congress on Blockchain and Applications

AU - Crisostomo, Joao

AU - Lobo, Victor

AU - Bacao, Fernando

N1 - Conference code: 5

PY - 2023/11/14

Y1 - 2023/11/14

N2 - The illicit activity in Blockchain reached an all-time high in 2021. In this work, we combined two machine learning techniques, Autoencoder (AE) and Extreme Gradient Boosting (XGBoost), to improve the performance of predicting illicit activity at the account level. The choice of autoencoding technique allows us to be able to detect new MOs (modus operandi) from fraudsters. With an Autoencoder trained only with healthy accounts, we are not misleading the model to focus on specific MOs as it happens with tree-based models. This allows us to introduce a dimension that could capture future fraudulent behaviours. Furthermore, the dataset was generated considering the real applicability of the model, i.e. it mimics what can realistically be obtained in a practical situation. With this approach, we were able to improve the state-of-the-art performance. In our test set the precision-recall AUC (area under the curve) of our final model increased by 12,07% when compared with our baseline.

AB - The illicit activity in Blockchain reached an all-time high in 2021. In this work, we combined two machine learning techniques, Autoencoder (AE) and Extreme Gradient Boosting (XGBoost), to improve the performance of predicting illicit activity at the account level. The choice of autoencoding technique allows us to be able to detect new MOs (modus operandi) from fraudsters. With an Autoencoder trained only with healthy accounts, we are not misleading the model to focus on specific MOs as it happens with tree-based models. This allows us to introduce a dimension that could capture future fraudulent behaviours. Furthermore, the dataset was generated considering the real applicability of the model, i.e. it mimics what can realistically be obtained in a practical situation. With this approach, we were able to improve the state-of-the-art performance. In our test set the precision-recall AUC (area under the curve) of our final model increased by 12,07% when compared with our baseline.

KW - Anomaly detection

KW - Ethereum

KW - Blockchain

KW - Autoencoder

KW - XGBoost

KW - Fraud Detection

KW - Malicious Accounts

KW - Machine Learning

U2 - 10.1007/978-3-031-45155-3_23

DO - 10.1007/978-3-031-45155-3_23

M3 - Conference contribution

SN - 978-3-031-45154-6

T3 - Lecture Notes in Networks and Systems

SP - 224

EP - 233

BT - Blockchain and Applications, 5th International Congress

A2 - Machado, José Manuel

A2 - Vieira, Paulo

A2 - Abelha, António

A2 - Vigneri, Luigi

A2 - Prieto, Javier

A2 - Peixoto, Hugo

A2 - Arroyo, David

PB - Springer, Cham

CY - Gewerbestrasse

Y2 - 12 July 2023 through 14 July 2023

ER -

Crisostomo J , Lobo V , Bacao F. Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost. In Machado JM, Vieira P, Abelha A, Vigneri L, Prieto J, Peixoto H, Arroyo D, editors, Blockchain and Applications, 5th International Congress: BLOCKCHAIN 2023. Gewerbestrasse: Springer, Cham. 2023. p. 224-233. (Lecture Notes in Networks and Systems). doi: 10.1007/978-3-031-45155-3_23

Detecting Fraudulent Wallets in Ethereum Blockchain Combining Supervised and Unsupervised Techniques: Using Autoencoders and XGboost

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Embargoed Document

Fingerprint

Cite this