Reproduce “Pricing Cross-Bank Context in Federated Anti-Money Laundering” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Lidia Losavio, Dario Fenoglio, Martin Gjoreski, Marc Langheinrich

The authors present Observable, which estimates a remote bank account’s graph-neural-network state from cross-bank transactions already recorded locally, without sending account-specific payloads. They compare it with sharing transaction counts or owner-computed states in AMLworld HI-Small partitions. At four silos, Observable’s pooled minority-class F1 gain over a baseline that drops cross-bank edges is .275; this unprotected benchmark does not establish real-bank effectiveness or end-to-end privacy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.