AMLNet: A Knowledge-Based Multi-Agent Framework to Generate and Detect Realistic Money Laundering Transactions

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Huda, Sabin, Foo, Ernest, Jadidi, Zahra, Newton, MA Hakim, Sattar, Abdul
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911155050512384
author Huda, Sabin
Foo, Ernest
Jadidi, Zahra
Newton, MA Hakim
Sattar, Abdul
author_facet Huda, Sabin
Foo, Ernest
Jadidi, Zahra
Newton, MA Hakim
Sattar, Abdul
contents Anti-money laundering (AML) research is constrained by the lack of publicly shareable, regulation-aligned transaction datasets. We present AMLNet, a knowledge-based multi-agent framework with two coordinated units: a regulation-aware transaction generator and an ensemble detection pipeline. The generator produces 1,090,173 synthetic transactions (approximately 0.16\% laundering-positive) spanning core laundering phases (placement, layering, integration) and advanced typologies (e.g., structuring, adaptive threshold behavior). Regulatory alignment reaches 75\% based on AUSTRAC rule coverage (Section 4.2), while a composite technical fidelity score of 0.75 summarizes temporal, structural, and behavioral realism components (Section 4.4). The detection ensemble achieves F1 0.90 (precision 0.84, recall 0.97) on the internal test partitions of AMLNet and adapts to the external SynthAML dataset, indicating architectural generalizability across different synthetic generation paradigms. We provide multi-dimensional evaluation (regulatory, temporal, network, behavioral) and release the dataset (Version 1.0, https://doi.org/10.5281/zenodo.16736515), to advance reproducible and regulation-conscious AML experimentation.
format Preprint
id arxiv_https___arxiv_org_abs_2509_11595
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AMLNet: A Knowledge-Based Multi-Agent Framework to Generate and Detect Realistic Money Laundering Transactions
Huda, Sabin
Foo, Ernest
Jadidi, Zahra
Newton, MA Hakim
Sattar, Abdul
Artificial Intelligence
Computational Engineering, Finance, and Science
Cryptography and Security
Machine Learning
Multiagent Systems
Anti-money laundering (AML) research is constrained by the lack of publicly shareable, regulation-aligned transaction datasets. We present AMLNet, a knowledge-based multi-agent framework with two coordinated units: a regulation-aware transaction generator and an ensemble detection pipeline. The generator produces 1,090,173 synthetic transactions (approximately 0.16\% laundering-positive) spanning core laundering phases (placement, layering, integration) and advanced typologies (e.g., structuring, adaptive threshold behavior). Regulatory alignment reaches 75\% based on AUSTRAC rule coverage (Section 4.2), while a composite technical fidelity score of 0.75 summarizes temporal, structural, and behavioral realism components (Section 4.4). The detection ensemble achieves F1 0.90 (precision 0.84, recall 0.97) on the internal test partitions of AMLNet and adapts to the external SynthAML dataset, indicating architectural generalizability across different synthetic generation paradigms. We provide multi-dimensional evaluation (regulatory, temporal, network, behavioral) and release the dataset (Version 1.0, https://doi.org/10.5281/zenodo.16736515), to advance reproducible and regulation-conscious AML experimentation.
title AMLNet: A Knowledge-Based Multi-Agent Framework to Generate and Detect Realistic Money Laundering Transactions
topic Artificial Intelligence
Computational Engineering, Finance, and Science
Cryptography and Security
Machine Learning
Multiagent Systems
url https://arxiv.org/abs/2509.11595