pith. sign in

arxiv: 2506.04292 · v3 · submitted 2025-06-04 · 💻 cs.SI · cs.LG· stat.AP

GARG-AML against Smurfing: A Scalable and Interpretable Graph-Based Framework for Anti-Money Laundering

Bruno Deprez , Bart Baesens , Tim Verdonck , Wouter Verbeke This is my paper

Pith reviewed 2026-05-19 11:25 UTC · model grok-4.3

classification 💻 cs.SI cs.LGstat.AP
keywords anti-money launderingsmurfinggraph-based detectionnetwork analysisfinancial fraudinterpretable modelsadjacency matrixsecond-order neighborhood
0
0 comments X

The pith

GARG-AML detects smurfing using density patterns in second-order neighborhood matrices

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

The paper develops GARG-AML to identify smurfing, a money laundering technique where small transactions are spread across many accounts to avoid detection. It does this by representing each account's immediate and secondary connections as an adjacency matrix and calculating densities in particular blocks of that matrix. These densities, along with basic network statistics, become features for simple classifiers like decision trees and gradient boosting. Tests on synthetic and open-source datasets show the method performs as well as or better than existing approaches. The key advantage is that it scales to very large transaction graphs typical in banks and produces risk scores that investigators can readily understand.

Core claim

GARG-AML demonstrates that smurfing patterns are captured by the densities of specific blocks in the adjacency matrix of an account's second-order neighborhood. This allows expert knowledge of laundering structures to be encoded directly into measurable network properties. Combined with standard classifiers, these properties generate interpretable risk scores that achieve state-of-the-art detection results on multiple datasets while remaining efficient for large-scale graphs.

What carries the argument

Second-order neighbourhood adjacency matrix whose block densities translate smurfing expert knowledge into risk-scoring features.

If this is right

  • Banks can deploy the framework on full-scale transaction networks without excessive computational demands.
  • Risk scores derive from transparent network measures, aiding investigators in reviewing and acting on alerts.
  • The method works for both directed and undirected graphs, fitting various transaction data formats.
  • Results remain competitive with complex models across synthetic and real-world open datasets.

Where Pith is reading between the lines

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

  • Neighborhood density features could be adapted to identify other forms of coordinated financial crime.
  • Adding time-based information to the matrix might improve detection of sequential laundering steps.
  • The lightweight nature supports embedding the scores into existing real-time monitoring pipelines at financial institutions.

Load-bearing premise

The density patterns within the second-order neighbourhood adjacency matrix reliably distinguish smurfing behavior in the tested datasets.

What would settle it

A test on an independent dataset with verified smurfing and legitimate accounts where the block density features do not yield separation in risk scores exceeding baseline random performance.

Figures

Figures reproduced from arXiv: 2506.04292 by Bart Baesens, Bruno Deprez, Tim Verdonck, Wouter Verbeke.

Figure 1
Figure 1. Figure 1: The two smurfing patterns described in the literature [ [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Example of a smurfing pattern. A is the node that will receive the score. The first-order neighbours (B, C, D) are shown in purple, and the second-order neighbour (E) is shown in green. A E B C D   0 0 1 1 1 0 0 1 1 1 1 1 0 0 0 1 1 0 0 0 1 1 0 0 0   (1) In general, the first row of the adjacency matrix, which corresponds to entries (A1,j ), represents the connections of node v under consideration… view at source ↗
Figure 3
Figure 3. Figure 3: The directed version of the smurfing pattern example, extended with an additional sending node (A’) and [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of a random network in which three different methods of injecting smurfing patterns are presented. [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The pipeline of the GARG-AML method. 11 [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Box plot of the run time for each method on [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: AUC-ROC and AUC-PR values of the compared methods across all synthetic datasets, with the split-on [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Ranks of the AUC-ROC and AUC-PR values of the compared methods across all synthetic datasets, with [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Critical distance plots of the average ranks [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 12
Figure 12. Figure 12: The AUC-ROC (solid bars) and AUC-PR (hatched bars) results on the HI-Small IBM dataset on the money [PITH_FULL_IMAGE:figures/full_fig_p016_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: The AUC-ROC (solid bars) and AUC-PR (hatched bars) results on the LI-Large IBM dataset on the money [PITH_FULL_IMAGE:figures/full_fig_p017_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Illustration of smurfing over different banks, which are represented by different colours. The (i)-icon [PITH_FULL_IMAGE:figures/full_fig_p018_14.png] view at source ↗
read the original abstract

Purpose: We introduce GARG-AML, a fast and transparent graph-based method to catch `smurfing', a common money-laundering tactic. It assigns a single, easy-to-understand risk score to every account in both directed and undirected networks. Unlike overly complex models, it balances detection power with the speed and clarity that investigators require. Methodology: The method maps an account's immediate and secondary connections (its second-order neighbourhood) into an adjacency matrix. By measuring the density of specific blocks within this matrix, GARG-AML flags patterns that mimic smurfing behaviour. We further boost the model's performance using decision trees and gradient-boosting classifiers, testing the results against current state-of-the-art on both synthetic and open-source data. Findings: GARG-AML matches or beats state-of-the-art performance across all tested datasets. Crucially, it easily processes the massive transaction graphs typical of large financial institutions. By leveraging only the adjacency matrix of the second-order neighbourhood and basic network features, this work highlights the potential of fundamental network properties towards advancing fraud detection. Originality: The originality lies in the translation of human expert knowledge of smurfing directly into a simple network representation, rather than relying on uninterpretable deep learning. Because GARG-AML is built expressly for the real-world business demands of scalability and interpretability, banks can easily incorporate it in their existing AML solutions.

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 / 2 minor

Summary. The paper introduces GARG-AML, a graph-based anti-money laundering framework that detects smurfing by mapping each account's second-order neighborhood into an adjacency matrix, computing densities of specific blocks within that matrix to encode smurfing patterns, and feeding these features plus basic network statistics into decision-tree and gradient-boosting classifiers. It reports that the approach matches or exceeds state-of-the-art performance on synthetic and open-source datasets while remaining scalable to large transaction graphs and more interpretable than deep-learning alternatives.

Significance. If the performance claims hold under rigorous validation, the work offers a practical, interpretable alternative to black-box models for AML, directly translating domain knowledge of smurfing into simple network-density features. The emphasis on scalability via adjacency-matrix operations and the avoidance of complex embeddings could be valuable for real-world deployment in financial institutions, provided the features demonstrably add signal beyond standard graph statistics.

major comments (2)
  1. [§4] §4 (Experimental Results) and Abstract: the claim that GARG-AML 'matches or beats state-of-the-art performance across all tested datasets' is unsupported by any reported numerical metrics, baseline names, validation protocol (e.g., train/test split, cross-validation), or error analysis. Without these, it is impossible to evaluate whether the reported gains are statistically meaningful or sensitive to data choices.
  2. [§3.2] §3.2 (Feature Construction): the central assumption that block densities in the second-order neighbourhood adjacency matrix reliably isolate smurfing signatures lacks supporting evidence such as ablation studies removing the density features, statistical comparison against degree-preserving null models, or explicit derivation of the chosen blocks from expert rules versus post-hoc tuning. In power-law transaction graphs this risks confounding with legitimate hub activity.
minor comments (2)
  1. [Abstract] The abstract and introduction repeatedly use 'second-order neighbourhood' without a precise definition or pseudocode for how the adjacency matrix is constructed from directed versus undirected edges.
  2. [Introduction] Missing references to prior graph-based AML work that also uses local density or motif features would help situate the novelty claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below and have revised the manuscript to provide the requested details and supporting analyses.

read point-by-point responses

  1. Referee: [§4] §4 (Experimental Results) and Abstract: the claim that GARG-AML 'matches or beats state-of-the-art performance across all tested datasets' is unsupported by any reported numerical metrics, baseline names, validation protocol (e.g., train/test split, cross-validation), or error analysis. Without these, it is impossible to evaluate whether the reported gains are statistically meaningful or sensitive to data choices.

    Authors: We agree that the performance claims require more explicit quantitative support. In the revised manuscript we will add a results table reporting concrete metrics (precision, recall, F1-score, AUC-ROC) for GARG-AML and the specific state-of-the-art baselines evaluated, together with the exact validation protocol (train/test splits and cross-validation folds) and a brief error analysis. These additions will allow direct assessment of statistical significance and sensitivity. revision: yes

  2. Referee: [§3.2] §3.2 (Feature Construction): the central assumption that block densities in the second-order neighbourhood adjacency matrix reliably isolate smurfing signatures lacks supporting evidence such as ablation studies removing the density features, statistical comparison against degree-preserving null models, or explicit derivation of the chosen blocks from expert rules versus post-hoc tuning. In power-law transaction graphs this risks confounding with legitimate hub activity.

    Authors: The selected blocks are derived directly from domain-expert descriptions of smurfing structures (multiple low-value transfers routed through intermediary accounts). In the revision we will add (i) an ablation study quantifying the contribution of the density features, (ii) a comparison against degree-preserving null models to demonstrate that the features capture more than degree distribution, and (iii) an explicit statement of the expert-rule origin of the block definitions. We will also clarify why the chosen blocks are unlikely to be confounded with generic hub activity in power-law graphs, as they target specific off-diagonal density patterns rather than overall degree. revision: yes

Circularity Check

0 steps flagged

No circularity: features from domain-expert block densities fed to standard classifiers

full rationale

The derivation constructs second-order neighbourhood adjacency matrices and extracts block densities as direct encodings of smurfing patterns drawn from expert knowledge. These engineered features are then passed to off-the-shelf decision trees and gradient-boosting classifiers. No equations reduce a claimed prediction to a fitted parameter by construction, no self-citation supplies a uniqueness theorem that forces the representation, and no ansatz is smuggled in. The central performance claims rest on empirical evaluation against external datasets rather than tautological re-use of inputs.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that smurfing produces distinguishable density signatures in second-order neighbourhoods; no explicit free parameters or invented entities are named in the abstract, though implementation thresholds are likely present.

free parameters (1)
  • Density block thresholds or cut-offs
    Likely used to define smurfing-like patterns, though not quantified in the abstract.
axioms (1)
  • domain assumption Smurfing behaviour manifests as identifiable density patterns in the second-order neighbourhood of transaction graphs.
    This premise directly converts human expert knowledge into the network representation used by GARG-AML.

pith-pipeline@v0.9.0 · 5808 in / 1230 out tokens · 59921 ms · 2026-05-19T11:25:40.126720+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

58 extracted references · 58 canonical work pages

  1. [1]

    URL https://data.world/lpetrocelli/czech-financial-dataset-real-anonymized-transactions

    1999. URL https://data.world/lpetrocelli/czech-financial-dataset-real-anonymized-transactions

    work page 1999

  2. [2]

    Competence of graph convolutional networks for anti-money laundering in bitcoin blockchain

    Ismail Alarab, Simant Prakoonwit, and Mohamed Ikbal Nacer. Competence of graph convolutional networks for anti-money laundering in bitcoin blockchain. In Proceedings of the 2020 5th International Conference on Machine Learning Technologies , ICMLT ’20, page 23–27, New York, NY , USA, 2020. Association for Computing Machinery. ISBN 9781450377645. doi:10.11...

    work page doi:10.1145/3409073.3409080 2020

  3. [3]

    Realistic synthetic financial transactions for anti-money laundering models

    Erik Altman, Jovan Blanuša, Luc von Niederhäusern, Beni Egressy, Andreea Anghel, and Kubilay Atasu. Realistic synthetic financial transactions for anti-money laundering models. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems, volume 36, pages 29851– 29874. Curran Associates...

    work page 2023

  4. [4]

    Emergence of scaling in random networks

    Albert-László Barabási and Réka Albert. Emergence of scaling in random networks. Science, 286(5439):509–512,

    work page

  5. [5]

    Fast unfolding of com- munities in large networks

    Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. Fast unfolding of com- munities in large networks. Journal of Statistical Mechanics: Theory and Experiment , 2008(10):P10008, oct

    work page 2008

  6. [6]

    Blondel, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre

    doi:10.1088/1742-5468/2008/10/P10008. URL https://dx.doi.org/10.1088/1742-5468/2008/10/ P10008

    work page doi:10.1088/1742-5468/2008/10/p10008 2008

  7. [7]

    Bolton and David J

    Richard J. Bolton and David J. Hand. Statistical Fraud Detection: A Review. Statistical Science, 17(3):235 – 255,

    work page

  8. [8]

    doi:10.1214/ss/1042727940

    work page doi:10.1214/ss/1042727940

  9. [9]

    Laundrograph: Self-supervised graph representation learning for anti-money laundering

    Mário Cardoso, Pedro Saleiro, and Pedro Bizarro. Laundrograph: Self-supervised graph representation learning for anti-money laundering. In Proceedings of the Third ACM International Conference on AI in Finance, ICAIF ’22, page 130–138, New York, NY , USA, 2022. Association for Computing Machinery. ISBN 9781450393768. doi:10.1145/3533271.3561727. URL https...

    work page doi:10.1145/3533271.3561727 2022

  10. [10]

    Machine learning techniques for anti-money laundering (aml) solutions in suspicious transaction detection: a review

    Zhiyuan Chen, Le Dinh Van Khoa, Ee Na Teoh, Amril Nazir, Ettikan Kandasamy Karuppiah, and Kim Sim Lam. Machine learning techniques for anti-money laundering (aml) solutions in suspicious transaction detection: a review. Knowledge and Information Systems, 57(2):245–285, 2018. doi:10.1007/s10115-017-1144-z. URL https://doi.org/10.1007/s10115-017-1144-z

    work page doi:10.1007/s10115-017-1144-z 2018

  11. [11]

    The relationship between precision-recall and roc curves

    Jesse Davis and Mark Goadrich. The relationship between precision-recall and roc curves. In Proceedings of the 23rd International Conference on Machine Learning , ICML ’06, page 233–240, New York, NY , USA,

    work page

  12. [12]

    ISBN 1595933832

    Association for Computing Machinery. ISBN 1595933832. doi:10.1145/1143844.1143874. URL https: //doi.org/10.1145/1143844.1143874

    work page doi:10.1145/1143844.1143874

  13. [13]

    Statistical comparisons of classifiers over multiple data sets

    Janez Demšar. Statistical comparisons of classifiers over multiple data sets. Journal of Machine Learning Research, 7(1):1–30, 2006. URL http://jmlr.org/papers/v7/demsar06a.html

    work page 2006

  14. [14]

    Network analyt- ics for anti-money laundering – a systematic literature review and experimental evaluation

    Bruno Deprez, Toon Vanderschueren, Bart Baesens, Tim Verdonck, and Wouter Verbeke. Network analyt- ics for anti-money laundering – a systematic literature review and experimental evaluation. arXiv preprint arXiv:2405.19383, 2024. URL https://arxiv.org/abs/2405.19383

    work page arXiv 2024

  15. [15]

    The application of social network analysis algorithms in a system supporting money laundering detection

    Rafał Dre˙zewski, Jan Sepielak, and Wojciech Filipkowski. The application of social network analysis algorithms in a system supporting money laundering detection. Information Sciences, 295:18–32, 2015. ISSN 0020-0255. doi:https://doi.org/10.1016/j.ins.2014.10.015. URL https://www.sciencedirect.com/science/article/ pii/S0020025514009979

    work page doi:10.1016/j.ins.2014.10.015 2015

  16. [16]

    Anomaly detection in graphs of bank transactions for anti money laundering applications

    Bogdan Dumitrescu, Andra B˘altoiu, and ¸ Stefania Budulan. Anomaly detection in graphs of bank transactions for anti money laundering applications. IEEE Access, 10:47699–47714, 2022. doi:10.1109/ACCESS.2022.3170467

    work page doi:10.1109/access.2022.3170467 2022

  17. [17]

    Provably powerful graph neural networks for directed multigraphs

    Béni Egressy, Luc von Niederhäusern, Jovan Blanuša, Erik Altman, Roger Wattenhofer, and Kubilay Atasu. Provably powerful graph neural networks for directed multigraphs. Proceedings of the AAAI Conference on Artificial Intelligence, 38(10):11838–11846, Mar. 2024. doi:10.1609/aaai.v38i10.29069. URL https://ojs. aaai.org/index.php/AAAI/article/view/29069

    work page doi:10.1609/aaai.v38i10.29069 2024

  18. [18]

    On random graphs

    Paul Erd ˝os and Alfréd Rényi. On random graphs. Publicationes Mathematicae Debrecen, 6(290-297):18, 1959

    work page 1959

  19. [19]

    International Standards on Combating Money Laundering and the Financing of Terrorism & Proliferation

    FATF. International Standards on Combating Money Laundering and the Financing of Terrorism & Proliferation. FATF, Paris, France, 2012-2023. URL www.fatf-gafi.org/en/publications/Fatfrecommendations/ Fatf-recommendations.html

    work page 2012

  20. [20]

    Resolution limit in community detection

    Santo Fortunato and Marc Barthélemy. Resolution limit in community detection. Proceedings of the National Academy of Sciences, 104(1):36–41, 2007. doi:10.1073/pnas.0605965104. URL https://www.pnas.org/doi/ abs/10.1073/pnas.0605965104. 19 GARG-AML against Smurfing: A Scalable and Interpretable Graph-Based Framework for Anti-Money Laundering

    work page doi:10.1073/pnas.0605965104 2007

  21. [21]

    Friedman, The use of ranks to avoid the assumption of normality implicit in the analysis of variance, Journal of the American Statistical Association 32 (1937) 675–701

    Milton Friedman. The use of ranks to avoid the assumption of normality implicit in the analysis of variance. Journal of the American Statistical Association, 32(200):675–701, 1937. doi:10.1080/01621459.1937.10503522. URL https://www.tandfonline.com/doi/abs/10.1080/01621459.1937.10503522

    work page doi:10.1080/01621459.1937.10503522 1937

  22. [22]

    A comparison of alternative tests of significance for the problem of m rankings

    Milton Friedman. A comparison of alternative tests of significance for the problem of m rankings. The Annals of Mathematical Statistics, 11(1):86–92, 1940. ISSN 00034851. URL http://www.jstor.org/stable/2235971

    work page arXiv 1940

  23. [23]

    Using social network analysis to prevent money laundering.Expert Systems with Applications, 67:49–58, 2017

    Andrea Fronzetti Colladon and Elisa Remondi. Using social network analysis to prevent money laundering.Expert Systems with Applications, 67:49–58, 2017. ISSN 0957-4174. doi:https://doi.org/10.1016/j.eswa.2016.09.029. URL https://www.sciencedirect.com/science/article/pii/S0957417416305139

    work page doi:10.1016/j.eswa.2016.09.029 2017

  24. [24]

    Schult, and Pieter J

    Aric Hagberg, Daniel A. Schult, and Pieter J. Swart. Exploring network structure, dynamics, and function using networkx. In Proceedings of the 7th Python in Science Conference (SciPy2008), page 11–15, 2008. URL https://www.osti.gov/biblio/960616

    work page 2008

  25. [25]

    Deep learning and explainable artificial intelligence techniques applied for detecting money laundering–a critical review

    Dattatray Vishnu Kute, Biswajeet Pradhan, Nagesh Shukla, and Abdullah Alamri. Deep learning and explainable artificial intelligence techniques applied for detecting money laundering–a critical review. IEEE Access, 9: 82300–82317, 2021

    work page 2021

  26. [26]

    Community detection algorithms: A comparative analysis

    Andrea Lancichinetti and Santo Fortunato. Community detection algorithms: A comparative analysis. Phys. Rev. E, 80:056117, 2009. doi:10.1103/PhysRevE.80.056117

    work page doi:10.1103/physreve.80.056117 2009

  27. [27]

    Autoaudit: Mining accounting and time-evolving graphs

    Meng-Chieh Lee, Yue Zhao, Aluna Wang, Pierre Jinghong Liang, Leman Akoglu, Vincent S Tseng, and Christos Faloutsos. Autoaudit: Mining accounting and time-evolving graphs. In 2020 IEEE International Conference on Big Data (Big Data), pages 950–956. IEEE, 2020. doi:10.1109/BigData50022.2020.9378346

    work page doi:10.1109/bigdata50022.2020.9378346 2020

  28. [28]

    Kevin J. Leonard. The development of a rule based expert system model for fraud alert in con- sumer credit. European Journal of Operational Research , 80(2):350–356, 1995. ISSN 0377-2217. doi:https://doi.org/10.1016/0377-2217(93)E0249-W. URL https://www.sciencedirect.com/science/ article/pii/0377221793E0249W

    work page doi:10.1016/0377-2217(93)e0249-w 1995

  29. [29]

    Money laundering

    Michael Levi and Peter Reuter. Money laundering. Crime and justice, 34(1):289–375, 2006. doi:10.1086/501508

    work page doi:10.1086/501508 2006

  30. [30]

    Flowscope: Spotting money laundering based on graphs

    Xiangfeng Li, Shenghua Liu, Zifeng Li, Xiaotian Han, Chuan Shi, Bryan Hooi, He Huang, and Xueqi Cheng. Flowscope: Spotting money laundering based on graphs. Proceedings of the AAAI Conference on Artificial Intelligence, 34(04):4731–4738, 2020. doi:10.1609/aaai.v34i04.5906. URL https://ojs.aaai.org/index. php/AAAI/article/view/5906

    work page doi:10.1609/aaai.v34i04.5906 2020

  31. [31]

    Knowledge discovery in cryptocurrency transactions: A survey

    Xiao Fan Liu, Xin-Jian Jiang, Si-Hao Liu, and Chi Kong Tse. Knowledge discovery in cryptocurrency transactions: A survey. Ieee access, 9:37229–37254, 2021. doi:10.1109/ACCESS.2021.3062652

    work page doi:10.1109/access.2021.3062652 2021

  32. [32]

    Kulatilleke, Mohanad Sarhan, Siamak Layeghy, and Marius Portmann

    Wai Weng Lo, Gayan K. Kulatilleke, Mohanad Sarhan, Siamak Layeghy, and Marius Portmann. Inspection-l: self- supervised gnn node embeddings for money laundering detection in bitcoin. Applied Intelligence, 53(16):19406– 19417, 2023. doi:10.1007/s10489-023-04504-9. URL https://doi.org/10.1007/s10489-023-04504-9

    work page doi:10.1007/s10489-023-04504-9 2023

  33. [33]

    Money laundering: a guide for criminal investigators

    John Madinger. Money laundering: a guide for criminal investigators. CRC Press, Boca Raton, 3rd ed. edition,

    work page

  34. [34]

    ISBN 978-1-4398-6912-3

    work page

  35. [35]

    Distribution-free multiple comparisons

    Peter Bjorn Nemenyi. Distribution-free multiple comparisons. Princeton University, 1963

    work page 1963

  36. [36]

    Finding and evaluating community structure in networks

    Mark EJ Newman and Michelle Girvan. Finding and evaluating community structure in networks. Physical review E, 69(2):026113, 2004. URL https://doi.org/10.1103/PhysRevE.69.026113

    work page doi:10.1103/physreve.69.026113 2004

  37. [37]

    The precision–recall curve overcame the optimism of the receiver operating characteristic curve in rare diseases

    Brice Ozenne, Fabien Subtil, and Delphine Maucort-Boulch. The precision–recall curve overcame the optimism of the receiver operating characteristic curve in rare diseases. Journal of clinical epidemiology, 68(8):855–859, 2015

    work page 2015

  38. [38]

    Transaction monitoring in anti-money laundering: A qualitative analysis and points of view from industry

    Berkan Oztas, Deniz Cetinkaya, Festus Adedoyin, Marcin Budka, Gokhan Aksu, and Huseyin Dogan. Transaction monitoring in anti-money laundering: A qualitative analysis and points of view from industry. Future Generation Computer Systems, 159:161–171, 2024

    work page 2024

  39. [39]

    Discovering bitcoin mixing using anomaly detection

    Mario Alfonso Prado-Romero, Christian Doerr, and Andrés Gago-Alonso. Discovering bitcoin mixing using anomaly detection. In Marcelo Mendoza and Sergio Velastín, editors, Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications, pages 534–541, Cham, 2018. Springer International Publishing. ISBN 978-3-319-75193-1

    work page 2018

  40. [40]

    Association rules applied to credit card fraud detection

    Daniel Sánchez, MA Vila, L Cerda, and José-Maria Serrano. Association rules applied to credit card fraud detection. Expert Systems with Applications, 36(2, Part 2):3630–3640, 2009. doi:https://doi.org/10.1016/j.eswa.2008.02.001

    work page doi:10.1016/j.eswa.2008.02.001 2009

  41. [41]

    Ted E Senator, Henry G Goldberg, Jerry Wooton, Matthew A Cottini, AF Umar Khan, Christina D Klinger, Winston M Llamas, Michael P Marrone, and Raphael WH Wong. Financial crimes enforcement network ai system 20 GARG-AML against Smurfing: A Scalable and Interpretable Graph-Based Framework for Anti-Money Laundering (fais) identifying potential money launderin...

    work page

  42. [42]

    doi:10.1609/aimag.v16i4.1169

    work page doi:10.1609/aimag.v16i4.1169

  43. [43]

    Smotef: Smurf money laundering detection using temporal order and flow analysis

    Shiva Shadrooh and Kjetil Nørvåg. Smotef: Smurf money laundering detection using temporal order and flow analysis. Applied Intelligence, 2024. doi:10.1007/s10489-024-05545-4. URL https://doi.org/10.1007/ s10489-024-05545-4

    work page doi:10.1007/s10489-024-05545-4 2024

  44. [44]

    Michele Starnini, Charalampos E. Tsourakakis, Maryam Zamanipour, André Panisson, Walter Allasia, Marco Fornasiero, Laura Li Puma, Valeria Ricci, Silvia Ronchiadin, Angela Ugrinoska, Marco Varetto, and Dario Moncalvo. Smurf-based anti-money laundering in time-evolving transaction networks. In Yuxiao Dong, Nicolas Kourtellis, Barbara Hammer, and Jose A. Loz...

    work page doi:10.1007/978-3-030-86514-6_11 2021

  45. [45]

    Topology-agnostic detection of temporal money laundering flows in billion- scale transactions, 2023

    Haseeb Tariq and Marwan Hassani. Topology-agnostic detection of temporal money laundering flows in billion- scale transactions, 2023

    work page 2023

  46. [46]

    V . A. Traag, L. Waltman, and N. J. van Eck. From louvain to leiden: guaranteeing well-connected communities. Scientific Reports, 9(1):5233, 2019. doi:10.1038/s41598-019-41695-z. URL https://doi.org/10.1038/ s41598-019-41695-z

    work page doi:10.1038/s41598-019-41695-z 2019

  47. [47]

    Money laundering

    United Nations Office on Drugs and Crime. Money laundering. https://www.unodc.org/unodc/en/ money-laundering/overview.html, (n.d.)

    work page

  48. [48]

    Catchm: A novel network-based credit card fraud detection method using node representation learning

    Rafaël Van Belle, Bart Baesens, and Jochen De Weerdt. Catchm: A novel network-based credit card fraud detection method using node representation learning. Decision Support Systems , 164:113866, 2023. ISSN 0167-9236. doi:https://doi.org/10.1016/j.dss.2022.113866

    work page doi:10.1016/j.dss.2022.113866 2023

  49. [49]

    Watts and Steven H

    Duncan J. Watts and Steven H. Strogatz. Collective dynamics of ‘small-world’networks. Nature, 393(6684): 440–442, 1998. doi:10.1038/30918. URL https://doi.org/10.1038/30918

    work page doi:10.1038/30918 1998

  50. [50]

    Weidele, Claudio Bellei, Tom Robinson, and Charles E

    Mark Weber, Giacomo Domeniconi, Jie Chen, Daniel Karl I. Weidele, Claudio Bellei, Tom Robinson, and Charles E. Leiserson. Anti-money laundering in bitcoin: Experimenting with graph convolutional networks for financial forensics, 2019

    work page 2019

  51. [51]

    Smurfs, money laundering, and the federal criminal law: the crime of structuring transactions

    Sarah N Welling. Smurfs, money laundering, and the federal criminal law: the crime of structuring transactions. Fla. L. Rev., 41:287, 1989

    work page 1989

  52. [52]

    Detecting mixing services via mining bitcoin transaction network with hybrid motifs

    Jiajing Wu, Jieli Liu, Weili Chen, Huawei Huang, Zibin Zheng, and Yan Zhang. Detecting mixing services via mining bitcoin transaction network with hybrid motifs. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 52(4):2237–2249, 2021. doi:10.1109/TSMC.2021.3049278

    work page doi:10.1109/tsmc.2021.3049278 2021

  53. [53]

    Towards understanding and demystifying bitcoin mixing services

    Lei Wu, Yufeng Hu, Yajin Zhou, Haoyu Wang, Xiapu Luo, Zhi Wang, Fan Zhang, and Kui Ren. Towards understanding and demystifying bitcoin mixing services. In Proceedings of the Web Conference 2021, WWW ’21, page 33–44, New York, NY , USA, 2021. Association for Computing Machinery. ISBN 9781450383127. doi:10.1145/3442381.3449880. URL https://doi.org/10.1145/3...

    work page doi:10.1145/3442381.3449880 2021

  54. [54]

    Distributed agents model for in- trusion detection based on ais

    Jin Yang, XiaoJie Liu, Tao Li, Gang Liang, and SunJun Liu. Distributed agents model for in- trusion detection based on ais. Knowledge-Based Systems , 22(2):115–119, 2009. ISSN 0950-

    work page 2009

  55. [55]

    URL https://www.sciencedirect.com/science/ article/pii/S0950705108001391

    doi:https://doi.org/10.1016/j.knosys.2008.07.005. URL https://www.sciencedirect.com/science/ article/pii/S0950705108001391

    work page doi:10.1016/j.knosys.2008.07.005 2008

  56. [56]

    Zhao Yang, René Algesheimer, and Claudio J. Tessone. A comparative analysis of community detection algorithms on artificial networks. Scientific Reports, 6(1):30750, 2016. doi:10.1038/srep30750. URL https: //doi.org/10.1038/srep30750

    work page doi:10.1038/srep30750 2016

  57. [57]

    No smurfs: Revealing fraud chains in mobile money transfers

    Maria Zhdanova, Jürgen Repp, Roland Rieke, Chrystel Gaber, and Baptiste Hemery. No smurfs: Revealing fraud chains in mobile money transfers. In 2014 Ninth International Conference on Availability, Reliability and Security, pages 11–20. IEEE, 2014. doi:10.1109/ARES.2014.10

    work page doi:10.1109/ares.2014.10 2014

  58. [58]

    Rule extraction from support vector machines based on con- sistent region covering reduction

    Pengfei Zhu and Qinghua Hu. Rule extraction from support vector machines based on con- sistent region covering reduction. Knowledge-Based Systems , 42:1–8, 2013. ISSN 0950-7051. doi:https://doi.org/10.1016/j.knosys.2012.12.003. URL https://www.sciencedirect.com/science/ article/pii/S095070511200336X. 21 GARG-AML against Smurfing: A Scalable and Interpreta...

    work page doi:10.1016/j.knosys.2012.12.003 2013