pith. sign in

arxiv: 2412.02408 · v3 · submitted 2024-12-03 · 💻 cs.SI · cs.LG· q-fin.GN

Leveraging Ensemble-Based Semi-Supervised Learning for Illicit Account Detection in Ethereum DeFi Transactions

Shabnam Fazliani , Mohammad Mowlavi Sorond , Arsalan Masoudifard This is my paper

Pith reviewed 2026-05-23 08:19 UTC · model grok-4.3

classification 💻 cs.SI cs.LGq-fin.GN
keywords illicit account detectionEthereumDeFisemi-supervised learningself-trainingIsolation Forestfraud detectionblockchain transactions
0
0 comments X

The pith

Self-learning ensemble detects illicit DeFi accounts on Ethereum with higher precision using less labeled data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes SLEID, a framework that starts with an Isolation Forest to flag outlier accounts and then applies self-training to generate pseudo-labels from unlabeled transaction data. This targets the problem of scarce labels when identifying illicit activity in Ethereum DeFi. A sympathetic reader would care because the method is tested on 6,903,860 transactions and reports gains of 2.56 percentage points in precision and 0.90 in F1 for the minority illicit class, plus 3.74 points in accuracy, while lowering the need for labeled examples. The approach treats the scarcity of ground-truth labels as the central constraint rather than assuming abundant supervision.

Core claim

SLEID uses an Isolation Forest model for initial outlier detection and a self-training mechanism to iteratively generate pseudo-labels for unlabeled accounts, enhancing detection accuracy. Experiments on 6,903,860 Ethereum transactions with extensive DeFi interaction coverage demonstrate that SLEID significantly outperforms supervised and semi-supervised baselines with +2.56 percentage-point precision, comparable recall, and +0.90 percentage-point F1 -- particularly for the minority illicit class -- alongside +3.74 percentage-points higher accuracy and improvements in PR-AUC, while substantially reducing reliance on labeled data.

What carries the argument

SLEID, the Self-Learning Ensemble-based Illicit account Detection framework, which integrates Isolation Forest for initial outlier identification and self-training to produce and refine pseudo-labels on unlabeled accounts.

If this is right

  • Detection performance improves especially on the minority illicit class without requiring additional labeled examples.
  • Accuracy rises by 3.74 percentage points and PR-AUC improves relative to both fully supervised and other semi-supervised baselines.
  • The framework lowers dependence on scarce ground-truth labels while maintaining comparable recall.
  • Iterative pseudo-labeling allows the model to exploit the large volume of unlabeled transaction data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same Isolation-Forest-plus-self-training pattern could be tested on transaction graphs from other blockchains to check whether the precision lift transfers.
  • If the pseudo-label quality holds, the method could be adapted for streaming detection where new transactions arrive continuously.
  • Pairing the ensemble outputs with simple graph features such as transaction degree or clustering coefficients might further stabilize the initial Isolation Forest step.

Load-bearing premise

The pseudo-labels created during self-training stay accurate enough across iterations to improve performance without spreading errors, and the Isolation Forest supplies a reliable initial separation of illicit accounts.

What would settle it

Retraining SLEID on a fresh collection of Ethereum DeFi transactions whose illicit labels have been independently verified, then measuring whether the reported gains in precision, F1, and accuracy disappear.

Figures

Figures reproduced from arXiv: 2412.02408 by Arsalan Masoudifard, Mohammad Mowlavi Sorond, Shabnam Fazliani.

Figure 1
Figure 1. Figure 1: Methodology overview for illicit account detection in the DeFi ecosystem. The pipeline consists of three main components: dataset preparation, model training, and prediction. Illicit ac￾counts are initially sourced from Etherscan and DeFi, followed by dataset expansion and refinement through network analysis. After vectorization and feature extraction, recursive feature elimination is applied to optimize t… view at source ↗
Figure 2
Figure 2. Figure 2: Network visualization of account interconnections. Blue, red, and gray nodes represent [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Performance Comparison of Illicit and Licit Class Detection Across Different Models [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Bipartite graph representation of our dataset. Sets A and B demonstrate the accounts [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of Isolation Forest contamination settings (0.25%, 0.5%, 1%) on downstream [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: 2D PCA Plot illustrating data dispersion, highlighting the need for a self-learning approach [PITH_FULL_IMAGE:figures/full_fig_p018_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Illicit Class Metrics over Self-Learning Iterations. This figure shows how precision, recall, [PITH_FULL_IMAGE:figures/full_fig_p019_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Local LIME explanation for a single account predicted as illicit. Red bars push toward [PITH_FULL_IMAGE:figures/full_fig_p020_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: SHAP summary plot showing global feature importance across the dataset. Trans [PITH_FULL_IMAGE:figures/full_fig_p021_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: XGBoost feature importance showing that transaction intensity, fee patterns, and network [PITH_FULL_IMAGE:figures/full_fig_p022_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Feature attributions for top false positives. Elevated activity and connectivity characteris [PITH_FULL_IMAGE:figures/full_fig_p022_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Imbalance-aware metrics for SLEID: PR-AUC (minority), MCC, and class-weighted F1. [PITH_FULL_IMAGE:figures/full_fig_p023_12.png] view at source ↗
read the original abstract

The advent of smart contracts has enabled the rapid rise of Decentralized Finance (DeFi) on the Ethereum blockchain, offering substantial rewards in financial innovation and inclusivity. This growth, however, is accompanied by significant security risks such as illicit accounts engaged in fraud. Effective detection is further limited by the scarcity of labeled data and the evolving tactics of malicious accounts. To address these challenges with a robust solution for safeguarding the DeFi ecosystem, we propose $\textbf{SLEID}$, a $\textbf{S}$elf-$\textbf{L}$earning $\textbf{E}$nsemble-based $\textbf{I}$llicit account $\textbf{D}$etection framework. SLEID uses an Isolation Forest model for initial outlier detection and a self-training mechanism to iteratively generate pseudo-labels for unlabeled accounts, enhancing detection accuracy. Experiments on 6,903,860 Ethereum transactions with extensive DeFi interaction coverage demonstrate that SLEID significantly outperforms supervised and semi-supervised baselines with $\textbf{+2.56}$ percentage-point precision, comparable recall, and $\textbf{+0.90}$ percentage-point F1 -- particularly for the minority illicit class -- alongside $\textbf{+3.74}$ percentage-points higher accuracy and improvements in PR-AUC, while substantially reducing reliance on labeled data.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes SLEID, a self-learning ensemble-based framework for illicit account detection in Ethereum DeFi transactions. It initializes with an Isolation Forest for outlier detection on unlabeled data and then applies iterative self-training to generate pseudo-labels, claiming substantial gains over supervised and semi-supervised baselines on 6,903,860 transactions: +2.56 pp precision, +0.90 pp F1 (especially on the minority illicit class), +3.74 pp accuracy, and improved PR-AUC, while reducing labeled-data requirements.

Significance. If the pseudo-labels prove reliable, the method could offer a practical way to exploit abundant unlabeled blockchain data for imbalanced detection tasks where labels are scarce and attack patterns evolve.

major comments (2)
  1. The self-training procedure is described without any validation of pseudo-label quality (e.g., confidence thresholding, disagreement filtering, or iterative held-out precision/recall measurement). This is load-bearing for the headline gains, as error amplification on the minority illicit class could inflate metrics relative to the supervised baselines.
  2. No information is given on data splitting, the exact proportion of labeled vs. unlabeled accounts, baseline hyper-parameter settings, or how the Isolation Forest seed labels were obtained and validated, preventing assessment of whether the reported improvements are robust.
minor comments (1)
  1. The abstract states numerical improvements but does not name the specific supervised and semi-supervised baselines or report the labeled-data fraction used in the experiments.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive feedback. We address each major comment below and will revise the manuscript accordingly to improve reproducibility and address concerns about pseudo-label reliability.

read point-by-point responses

  1. Referee: The self-training procedure is described without any validation of pseudo-label quality (e.g., confidence thresholding, disagreement filtering, or iterative held-out precision/recall measurement). This is load-bearing for the headline gains, as error amplification on the minority illicit class could inflate metrics relative to the supervised baselines.

    Authors: We acknowledge that the current description of the self-training procedure lacks explicit validation steps for the generated pseudo-labels. We will revise the manuscript to include details on the confidence thresholding applied during iterative self-training, along with iterative held-out precision and recall measurements on a validation set. This addition will demonstrate the stability of the pseudo-labels and directly address potential error amplification risks for the minority illicit class. revision: yes

  2. Referee: No information is given on data splitting, the exact proportion of labeled vs. unlabeled accounts, baseline hyper-parameter settings, or how the Isolation Forest seed labels were obtained and validated, preventing assessment of whether the reported improvements are robust.

    Authors: We agree that these experimental details are necessary for assessing robustness and reproducibility. The revised manuscript will add a dedicated experimental setup subsection specifying the data splitting strategy and proportions, the exact labeled-to-unlabeled account ratio, hyperparameter values for the Isolation Forest and all baselines, and the procedure used to obtain and validate the initial Isolation Forest seed labels. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical method with standard semi-supervised components

full rationale

The paper introduces SLEID as an ensemble self-training framework using Isolation Forest for initial labels and iterative pseudo-labeling, then reports empirical gains on a fixed dataset of 6.9M transactions against supervised and semi-supervised baselines. No equations, fitted parameters, or uniqueness claims are shown that reduce by construction to the inputs (no self-definitional loops, no predictions that are statistically forced by the fit itself, and no load-bearing self-citations). Self-training is a recognized technique whose risks are external to the derivation; the reported metrics are measured on held-out data rather than tautologically on the pseudo-labels. The chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Relies on common assumptions in semi-supervised outlier detection without introducing new entities or parameters explicitly.

axioms (1)
  • domain assumption Self-training iteratively improves detection by generating reliable pseudo-labels from confident predictions.
    This is the core mechanism of the self-learning component.

pith-pipeline@v0.9.0 · 5773 in / 1262 out tokens · 34872 ms · 2026-05-23T08:19:40.117088+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

51 extracted references · 51 canonical work pages · 1 internal anchor

  1. [1]

    Controlling the latent space of gans through reinforcement learning: A case study on task-based image-to-image translation

    Mahyar Abbasian, Taha Rajabzadeh, Ahmadreza Moradipari, Seyed Amir Hossein Aqajari, Hongsheng Lu, and Amir M Rahmani. Controlling the latent space of gans through reinforcement learning: A case study on task-based image-to-image translation. In Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing, pp.\ 1061--1063, 2024

    work page 2024

  2. [2]

    Lgbm: a machine learning approach for ethereum fraud detection

    Rabia Musheer Aziz, Mohammed Farhan Baluch, Sarthak Patel, and Abdul Hamid Ganie. Lgbm: a machine learning approach for ethereum fraud detection. International Journal of Information Technology, 14 0 (7): 0 3321--3331, 2022 a

    work page 2022

  3. [3]

    A machine learning based approach to detect the ethereum fraud transactions with limited attributes

    Rabia Musheer Aziz, Mohammed Farhan Baluch, Sarthak Patel, and Pavan Kumar. A machine learning based approach to detect the ethereum fraud transactions with limited attributes. Karbala International Journal of Modern Science, 8 0 (2): 0 139--151, 2022 b

    work page 2022

  4. [4]

    Modified genetic algorithm with deep learning for fraud transactions of ethereum smart contract

    Rabia Musheer Aziz, Rajul Mahto, Kartik Goel, Aryan Das, Pavan Kumar, and Akash Saxena. Modified genetic algorithm with deep learning for fraud transactions of ethereum smart contract. Applied Sciences, 13 0 (2): 0 697, 2023

    work page 2023

  5. [5]

    Money laundering activity spread across more service deposit addresses in 2023, plus new tactics from lazarus group

    Chainalysis . Money laundering activity spread across more service deposit addresses in 2023, plus new tactics from lazarus group. https://www.chainalysis.com/blog/2024-crypto-money-laundering/, February 2024

    work page 2023

  6. [6]

    Anti-money laundering by group-aware deep graph learning

    Dawei Cheng, Yujia Ye, Sheng Xiang, Zhenwei Ma, Ying Zhang, and Changjun Jiang. Anti-money laundering by group-aware deep graph learning. IEEE Trans. on Knowledge and Data Engineering, 35 0 (12): 0 12444--12456, 2023

    work page 2023

  7. [7]

    Detection of illicit accounts over the ethereum blockchain

    Simone Farrugia et al. Detection of illicit accounts over the ethereum blockchain. Expert Systems with Applications, pp.\ 113318, 2020. doi:10.1016/j.eswa.2020.113318

    work page doi:10.1016/j.eswa.2020.113318 2020

  8. [8]

    Hausdorff measure bound for the nodal sets of neumann laplace eigenfunctions, 2024

    Shaghayegh Fazliani. Hausdorff measure bound for the nodal sets of neumann laplace eigenfunctions, 2024. URL https://arxiv.org/abs/2311.09686

    work page arXiv 2024

  9. [9]

    Enhancing physics-informed neural networks through feature engineering, 2025

    Shaghayegh Fazliani, Zachary Frangella, and Madeleine Udell. Enhancing physics-informed neural networks through feature engineering, 2025. URL https://arxiv.org/abs/2502.07209

    work page arXiv 2025

  10. [11]

    A general framework for account risk rating on ethereum: toward safer blockchain technology

    Qishuang Fu, Dan Lin, Jiajing Wu, and Zibin Zheng. A general framework for account risk rating on ethereum: toward safer blockchain technology. IEEE Transactions on Computational Social Systems, 11 0 (2): 0 1865--1875, 2023 b

    work page 2023

  11. [12]

    Vw-dbg: A dynamically evolving bitcoin transaction network model

    Jinke Geng, Yi Li, Li Fang, and Ping Chen. Vw-dbg: A dynamically evolving bitcoin transaction network model. IEEE Transactions on Network Science and Engineering, 9 0 (2): 0 356--363, 2021

    work page 2021

  12. [13]

    Anti-money laundering in cryptocurrency via multi-relational graph neural network

    Woochang Hyun, Jaehong Lee, and Bongwon Suh. Anti-money laundering in cryptocurrency via multi-relational graph neural network. In PAKDD, LNAI 13936, pp.\ 118--130, 2023

    work page 2023

  13. [14]

    Fighting money laundering with statistics and machine learning

    Rasmus Ingemann Tuffveson Jensen and Alexandros Iosifidis. Fighting money laundering with statistics and machine learning. IEEE Access, 11: 0 8889--8902, 2023

    work page 2023

  14. [15]

    Ethereum fraud detection with heterogeneous graph neural networks

    Hiroki Kanezashi, Toyotaro Suzumura, Xin Liu, and Takahiro Hirofuchi. Ethereum fraud detection with heterogeneous graph neural networks. In Conference Proceedings, 2018

    work page 2018

  15. [16]

    Labanca, L

    D. Labanca, L. Primerano, M. Markland-Montgomery, M. Polino, M. Carminati, and S. Zanero. Amaretto: An active learning framework for money laundering detection. IEEE Access, 10: 0 41720--41732, 2022

    work page 2022

  16. [17]

    Strong dispersive coupling between a mechanical resonator and a fluxonium superconducting qubit

    Nathan RA Lee, Yudan Guo, Agnetta Y Cleland, E Alex Wollack, Rachel G Gruenke, Takuma Makihara, Zhaoyou Wang, Taha Rajabzadeh, Wentao Jiang, Felix M Mayor, et al. Strong dispersive coupling between a mechanical resonator and a fluxonium superconducting qubit. PRX Quantum, 4 0 (4): 0 040342, 2023

    work page 2023

  17. [18]

    Siege: Self-supervised incremental deep graph learning for ethereum phishing scam detection

    Shuo Li et al. Siege: Self-supervised incremental deep graph learning for ethereum phishing scam detection. 2023. doi:10.1145/3581783.3612461

    work page doi:10.1145/3581783.3612461 2023

  18. [19]

    Graph embedding-based money laundering detection for ethereum

    Jiayi Liu, Changchun Yin, Hao Wang, Xiaofei Wu, Dongwan Lan, Lu Zhou, and Chunpeng Ge. Graph embedding-based money laundering detection for ethereum. Electronics, 12 0 (14): 0 3180, 2023. doi:10.3390/electronics12143180

    work page doi:10.3390/electronics12143180 2023

  19. [20]

    Leveraging graph-rag and prompt engineering to enhance llm-based automated requirement traceability and compliance checks

    Arsalan Masoudifard, Mohammad Mowlavi Sorond, Moein Madadi, Mohammad Sabokrou, and Elahe Habibi. Leveraging graph-rag and prompt engineering to enhance llm-based automated requirement traceability and compliance checks. arXiv preprint arXiv:2412.08593, 2024

    work page arXiv 2024

  20. [21]

    Multi-environment meta-learning in stochastic linear bandits

    Ahmadreza Moradipari, Mohammad Ghavamzadeh, Taha Rajabzadeh, Christos Thrampoulidis, and Mahnoosh Alizadeh. Multi-environment meta-learning in stochastic linear bandits. In 2022 IEEE International Symposium on Information Theory (ISIT), pp.\ 1659--1664. IEEE, 2022

    work page 2022

  21. [22]

    Cybersecurity anomaly detection: Ai and ethereum blockchain for a secure and tamperproof ioht data management

    Oluwaseun Priscilla Olawale and Sahar Ebadinezhad. Cybersecurity anomaly detection: Ai and ethereum blockchain for a secure and tamperproof ioht data management. IEEE Access, 2024

    work page 2024

  22. [23]

    Bitcoin money laundering detection via subgraph contrastive learning

    Shiyu Ouyang, Qianlan Bai, Hui Feng, and Bo Hu. Bitcoin money laundering detection via subgraph contrastive learning. Entropy, 26 0 (3): 0 211, 2024

    work page 2024

  23. [24]

    Leveraging machine learning for multichain defi fraud detection

    Georgios Palaiokrassas et al. Leveraging machine learning for multichain defi fraud detection. ArXiv.org, 2023. Available at: https://arxiv.org/abs/2306.07972

    work page arXiv 2023

  24. [25]

    Detecting malicious ethereum entities via application of machine learning classification

    Farimah Poursafaei, Ghaith Bany Hamad, and Zeljko Zilic. Detecting malicious ethereum entities via application of machine learning classification. In 2020 2nd conference on blockchain research & applications for innovative networks and services (BRAINS), pp.\ 120--127. IEEE, 2020

    work page 2020

  25. [26]

    An empirical study of DeFi liquidations: incentives, risks, and instabilities

    Kaihua Qin, Liyi Zhou, Pablo Gau, Peter Jovanovic, and Arthur Gervais. An empirical study of defi liquidations: Incentives, risks, and instabilities. In Proceedings of the 2021 ACM Conference on Economics and Computation, pp.\ 1--18, 2021. doi:10.1145/3487552.3487811

    work page doi:10.1145/3487552.3487811 2021

  26. [27]

    A deep learning model for threat hunting in ethereum blockchain

    Elnaz Rabieinejad, Abbas Yazdinejad, and Reza M Parizi. A deep learning model for threat hunting in ethereum blockchain. In 2021 IEEE 20th International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom), pp.\ 1185--1190. IEEE, 2021

    work page 2021

  27. [28]

    Femtosecond CDMA Using Dielectric Metasurfaces: Design Procedure and Challenges

    Taha Rajabzadeh, Mohammad Hosein Mousavi, Sajjad Abdollahramezani, Mohammad Vahid Jamali, and Jawad A Salehi. Femtosecond cdma using dielectric metasurfaces: Design procedure and challenges. arXiv preprint arXiv:1712.00834, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  28. [29]

    Photonics-to-free-space interface in lithium niobate-on-sapphire

    Taha Rajabzadeh, Christopher J Sarabalis, Okan Atalar, and Amir H Safavi-Naeini. Photonics-to-free-space interface in lithium niobate-on-sapphire. In CLEO: Science and Innovations, pp.\ STu4J--6. Optica Publishing Group, 2020

    work page 2020

  29. [30]

    Analysis of arbitrary superconducting quantum circuits accompanied by a python package: Sqcircuit

    Taha Rajabzadeh, Zhaoyou Wang, Nathan Lee, Takuma Makihara, Yudan Guo, and Amir H Safavi-Naeini. Analysis of arbitrary superconducting quantum circuits accompanied by a python package: Sqcircuit. Quantum, 7: 0 1118, 2023

    work page 2023

  30. [31]

    A general framework for gradient-based optimization of superconducting quantum circuits using qubit discovery as a case study

    Taha Rajabzadeh, Alex Boulton-McKeehan, Sam Bonkowsky, David I Schuster, and Amir H Safavi-Naeini. A general framework for gradient-based optimization of superconducting quantum circuits using qubit discovery as a case study. arXiv preprint arXiv:2408.12704, 2024 a

    work page arXiv 2024

  31. [32]

    Gradient-based optimization of superconducting quantum circuit designs-part 2

    Taha Rajabzadeh, Alexander Boulton-McKeehan, Sam Bonkowsky, and Amir Safavi-Naeini. Gradient-based optimization of superconducting quantum circuit designs-part 2. In APS March Meeting Abstracts, volume 2024, pp.\ A47--010, 2024 b

    work page 2024

  32. [33]

    Turbocharging gaussian process inference with approximate sketch-and-project, 2025

    Pratik Rathore, Zachary Frangella, Sachin Garg, Shaghayegh Fazliani, Michał Dereziński, and Madeleine Udell. Turbocharging gaussian process inference with approximate sketch-and-project, 2025. URL https://arxiv.org/abs/2505.13723

    work page arXiv 2025

  33. [34]

    Sarker, Md Musfique Anwar, Md

    Rony Chowdhury Ripan, Iqbal H. Sarker, Md Musfique Anwar, Md. Hasan Furhad, Fazle Rahat, Mohammed Moshiul Hoque, and Muhammad Sarfraz. An isolation forest learning based outlier detection approach for effectively classifying cyber anomalies. In 2021 International Conference on Data Science and Security (ICDSS), 2021. doi:10.1007/springer-12345

    work page doi:10.1007/springer-12345 2021

  34. [35]

    Abnormal transactions detection in the ethereum network using semi-supervised generative adversarial networks

    Yousef K Sanjalawe and Salam R Al-E’mari. Abnormal transactions detection in the ethereum network using semi-supervised generative adversarial networks. IEEE Access, 2023

    work page 2023

  35. [36]

    Decentralized finance: On blockchain- and smart contract-based financial markets

    Fabian Schär. Decentralized finance: On blockchain- and smart contract-based financial markets. Papers.ssrn.com, 2021. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3843844

    work page 2021

  36. [37]

    Semi-supervised learning for anomaly detection in blockchain-based supply chains

    Do Hai Son, Bui Duc Manh, Tran Viet Khoa, Nguyen Linh Trung, Dinh Thai Hoang, Hoang Trong Minh, Yibeltal Alem, and Le Quang Minh. Semi-supervised learning for anomaly detection in blockchain-based supply chains. In 2024 23rd International Symposium on Communications and Information Technologies (ISCIT), pp.\ 140--145. IEEE, 2024

    work page 2024

  37. [38]

    Adaptive attention-based graph representation learning to detect phishing accounts on the ethereum blockchain

    Haojie Sun. Adaptive attention-based graph representation learning to detect phishing accounts on the ethereum blockchain. IEEE Transactions on Network Science and Engineering, pp.\ --, 2024. doi:10.1109/TNSE.2024.3355089

    work page doi:10.1109/tnse.2024.3355089 2024

  38. [39]

    Ethereum fraud behavior detection based on graph neural networks

    Runnan Tan, Qingfeng Tan, Qin Zhang, Peng Zhang, Yushun Xie, and Zhao Li. Ethereum fraud behavior detection based on graph neural networks. Computing, 105 0 (10): 0 2143--2170, 2023

    work page 2023

  39. [40]

    Ensemble deep learning based prediction of fraudulent cryptocurrency transactions

    Qasim Umer, Jian-Wei Li, Muhammad Rehan Ashraf, Rab Nawaz Bashir, and Hamid Ghous. Ensemble deep learning based prediction of fraudulent cryptocurrency transactions. IEEE Access, 2023

    work page 2023

  40. [41]

    Department of the Treasury

    U.S. Department of the Treasury . Defi illicit finance risk assessment, 2023. Available at: https://home.treasury.gov/news/press-releases/jy1391

    work page 2023

  41. [42]

    Graph deep learning based anomaly detection in ethereum blockchain network

    Lei Pan Vatsal Patel and Sutharshan Rajasegarar. Graph deep learning based anomaly detection in ethereum blockchain network. In International Conference on Network and System Security (NSS), pp.\ 132--148. Springer, 2020. doi:10.1007/978-3-030-65745-1_8

    work page doi:10.1007/978-3-030-65745-1_8 2020

  42. [43]

    Who are the phishers? phishing scam detection on ethereum via network embedding

    Jiajing Wu, Qi Yuan, Dan Lin, Weijia You, Wei Chen, Chen Chen, and Zibin Zheng. Who are the phishers? phishing scam detection on ethereum via network embedding. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 52 0 (2): 0 1156--1166, 2020. doi:10.1109/TSMC.2020.2979892

    work page doi:10.1109/tsmc.2020.2979892 2020

  43. [44]

    Analysis of cryptocurrency transactions from a network perspective: An overview

    Jiajing Wu, Jieli Liu, Yijing Zhao, and Zibin Zheng. Analysis of cryptocurrency transactions from a network perspective: An overview. Journal of Network and Computer Applications, 190: 0 103139, 2021

    work page 2021

  44. [45]

    Toward understanding asset flows in crypto money laundering through the lenses of ethereum heists

    Jiajing Wu, Dan Lin, Qishuang Fu, Shuo Yang, Ting Chen, Zibin Zheng, and Bowen Song. Toward understanding asset flows in crypto money laundering through the lenses of ethereum heists. IEEE Trans. on Information Forensics and Security, 19: 0 1994--2005, 2024 a

    work page 1994

  45. [46]

    Toward understanding asset flows in crypto money laundering through the lenses of ethereum heists

    Jiawei Wu, Dongze Lin, Qian Fu, Sheng Yang, Tong Chen, Zibin Zheng, and Bo Song. Toward understanding asset flows in crypto money laundering through the lenses of ethereum heists. IEEE Transactions on Information Forensics and Security, 19: 0 1994--2007, 2024 b . doi:10.1109/TIFS.2023.3346276

    work page doi:10.1109/tifs.2023.3346276 1994

  46. [47]

    Visual analysis of money laundering in cryptocurrency exchange

    Fangfang Zhou, Yunpeng Chen, Chunyao Zhu, Lijia Jiang, Xincheng Liao, Zengsheng Zhong, Xiaohui Chen, Yi Chen, and Ying Zhao. Visual analysis of money laundering in cryptocurrency exchange. IEEE Trans. on Computational Social Systems, 11 0 (1): 0 731--742, 2024

    work page 2024

  47. [48]

    Sok: Decentralized finance (defi) attacks

    Liyi Zhou, Xihan Xiong, Jens Ernstberger, Stefanos Chaliasos, Zhipeng Wang, Ye Wang, Kaihua Qin, Roger Wattenhofer, Dawn Song, and Arthur Gervais. Sok: Decentralized finance (defi) attacks. pp.\ 2444--2461, 2023

    work page 2023

  48. [49]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

    work page

  49. [50]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should not add it explicitly Type <Return> for now, but then later remove the command n...

    work page

  50. [51]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@first@sw \@firstoftwo \@ifundefined NAT@b*@#2 \@firstoftwo @num @NAT@ctr \@secondoft...

    work page

  51. [52]

    adobe:ns:meta/

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibsetup #1 @NAT@ctr @ @openbib .11em \@plus.33em \@minus.07em 4000 4000 `\.\@m @bibit...

    work page 2024