Tide: A Customisable Dataset Generator for Anti-Money Laundering Research

Montijn van den Beukel, Jože M. Rožanec, Ana-Lucia Varbanescu · 2026

Read Tide: A Customisable Dataset Generator for Anti-Money Laundering Research by Montijn van den Beukel, Jože M. Rožanec, Ana-Lucia Varbanescu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-01863.txt.

Description

Montijn van den Beukel, Joze Rozanec, and Ana-Lucia Varbanescu build fake bank data to train money laundering detectors. Their 2026 work starts from a block: privacy and law keep real transaction records out of research reach. Past fake data sets copy simple structure but miss the timing patterns of real laundering schemes.

Their open generator, Tide, builds graph-based transaction networks with structure and timing both modeled. Users can tune the output to their research needs and reproduce it exactly. Two reference sets ship with the tool, at low and higher illicit shares, plus top detector models. Rankings flip between the two sets: one gradient method leads at low illicit share, another at higher share. That flip proves the benchmark can separate models by operating conditions.

Catalog record

CREATORMontijn van den Beukel, Jože M. Rožanec, Ana-Lucia Varbanescu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE97198
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.01863

Browse the full library · About the Tower of Babel